Merge branch 'tf2' into hl2dm

# Conflicts:
#	game/client/cdll_util.cpp
#	game/client/clientleafsystem.cpp
#	game/client/clientshadowmgr.cpp
#	game/client/detailobjectsystem.cpp
#	game/server/filters.cpp
#	game/server/hltvdirector.cpp
#	game/server/physconstraint.cpp
#	game/server/props.cpp
#	game/shared/gamerules.cpp
#	game/shared/takedamageinfo.cpp
#	game/shared/takedamageinfo.h
#	lib/linux/tier1_i486.a
#	lib/mac/tier1_i486.a
#	lib/public/tier1.lib
#	public/eiface.h
#	public/mathlib/mathlib.h
#	public/tier0/memalloc.h
#	public/tier1/utlmemory.h
#	public/tier1/utlvector.h
#	public/toolframework/itoolentity.h
#	tier1/KeyValues.cpp

AMBuildScript

@@ -0,0 +1,161 @@
+# vim: set ts=2 sw=2 tw=99 noet ft=python: 
+import os, sys, shutil
+from ambuild2 import util
+
+def ResolveEnvPath(env, folder=None):
+  if env in os.environ:
+    path = os.environ[env]
+    if os.path.isdir(path):
+      return path
+    return None
+
+  if folder:
+    head = os.getcwd()
+    oldhead = None
+    while head != None and head != oldhead:
+      path = os.path.join(head, folder)
+      if os.path.isdir(path):
+        return path
+      oldhead = head
+      head, tail = os.path.split(head)
+
+  return None
+
+def Normalize(path):
+  return os.path.abspath(os.path.normpath(path))
+
+class SDKConfig(object):
+  def __init__(self):
+    self.libs = []
+    self.targets = []
+    self.target_archs = set()
+
+    if builder.options.targets:
+      target_archs = builder.options.targets.split(',')
+    else:
+      target_archs = ['x86', 'x86_64']
+
+    for arch in target_archs:
+      try:
+        cxx = builder.DetectCxx(target_arch = arch)
+        self.target_archs.add(cxx.target.arch)
+      except Exception as e:
+        if builder.options.targets:
+            raise
+        print('Skipping target {}: {}'.format(arch, e))
+        continue
+      self.targets.append(cxx)
+
+    if not self.targets:
+      raise Exception('No suitable C/C++ compiler was found.')
+  
+  @property
+  def tag(self):
+    if builder.options.debug == '1':
+      return 'Debug'
+    return 'Release'
+  
+  def configure_cxx(self, cxx):
+    if cxx.like('gcc'):
+      self.configure_gcc(cxx)
+    elif cxx.family == 'msvc':
+      self.configure_msvc(cxx)
+
+    # Optimization
+    if builder.options.opt == '1':
+      cxx.defines += ['NDEBUG']
+
+    # Debugging
+    if builder.options.debug == '1':
+      cxx.defines += ['DEBUG', '_DEBUG']
+
+    if cxx.target.arch == 'x86_64':
+      cxx.defines += ['X64BITS', 'PLATFORM_64BITS']
+
+    # Platform-specifics
+    if cxx.target.platform == 'linux':
+      self.configure_linux(cxx)
+    elif cxx.target.platform == 'windows':
+      self.configure_windows(cxx)
+  
+  def configure_gcc(self, cxx):
+    cxx.cflags += [
+      '-Wall',
+      '-Werror',
+      '-Wno-overloaded-virtual',
+      '-msse',
+      '-fPIC'
+    ]
+
+    cxx.defines += [
+      'COMPILER_GCC'
+    ]
+
+    cxx.cxxflags += [
+      '-std=c++17'
+    ]
+
+    if builder.options.opt == '1':
+      cxx.cflags += ['-O3']
+    return
+
+  def configure_msvc(self, cxx):
+    cxx.defines += ['COMPILER_MSVC']
+    if cxx.target.arch == 'x86':
+      cxx.defines += ['COMPILER_MSVC32']
+    elif cxx.target.arch == 'x86_64':
+      cxx.defines += ['COMPILER_MSVC64']
+    
+    cxx.cxxflags += [
+      '/std:c++17',
+      '/WX'
+    ]
+    
+    if builder.options.opt == '1':
+      cxx.cflags += ['/Ox', '/Zo']
+      cxx.linkflags += ['/OPT:ICF', '/OPT:REF']
+
+    return
+  
+  def configure_linux(self, cxx):
+    cxx.defines += ['_LINUX', 'LINUX', 'POSIX', 'GNUC']
+
+    # Set of defines required by the HL2SDK
+    cxx.defines += [
+      '_finite=finite', 'stricmp=strcasecmp',
+      '_stricmp=strcasecmp', '_strnicmp=strncasecmp',
+	    'strnicmp=strncasecmp', '_vsnprintf=vsnprintf',
+      '_alloca=alloca', 'strcmpi=strcasecmp',
+      'NO_MALLOC_OVERRIDE'
+    ]
+    return
+  
+  def configure_windows(self, cxx):
+    cxx.defines += ['_WINDOWS']
+    if cxx.target.arch == 'x86':
+      cxx.defines += ['WIN32']
+    elif cxx.target.arch == 'x86_64':
+      cxx.defines += ['WIN64']
+    return
+  
+  def configure(self):
+    for cxx in self.targets:
+      self.configure_cxx(cxx)
+
+  def ConfigureLibrary(self, project, compiler, context, prefix = ''):
+    name = prefix + project.name
+    if compiler.target.platform == 'linux' and compiler.target.arch == 'x86':
+      name += '_i486'
+    binary = project.Configure(compiler, name, '{0} - {1}'.format(self.tag, compiler.target.arch))
+    binary.compiler.cxxincludes += [
+      os.path.join(context.currentSourcePath)
+    ]
+    return binary
+
+HL2SDK = SDKConfig()
+HL2SDK.configure()
+
+BuildScripts = ['mathlib/AMBuilder', 'tier1/AMBuilder', 'PackageScript']
+# Turn off the 'lib' prefix for all platform, we will add it if we need to
+util.StaticLibPrefix = ''
+builder.Build(BuildScripts, { 'HL2SDK': HL2SDK })

PackageScript

@@ -0,0 +1,56 @@
+# vim: set ts=8 sts=2 sw=2 tw=99 et ft=python:
+import os
+
+# This is where the files will be output to
+# package is the default
+builder.SetBuildFolder('lib')
+
+
+folder_list = []
+
+for task in HL2SDK.libs:
+  if task.target.platform == 'windows':
+    folder_list += ['public', 'public/win64']
+    break
+
+for task in HL2SDK.libs:
+  if task.target.platform == 'linux':
+    folder_list += ['linux', 'linux64']
+    break
+
+# Create the distribution folder hierarchy.
+folder_map = {}
+for folder in folder_list:
+  norm_folder = os.path.normpath(folder)
+  folder_map[folder] = builder.AddFolder(norm_folder)
+
+# Do all straight-up file copies from the source tree.
+def CopyFiles(src, dest, files):
+  if not dest:
+    dest = src
+  dest_entry = folder_map[dest]
+  for source_file in files:
+    source_path = os.path.join(builder.sourcePath, src, source_file)
+    builder.AddCopy(source_path, dest_entry)
+def CopyFile(src, dest):
+  dest_entry = folder_map[dest]
+  source_path = os.path.join(builder.sourcePath, src)
+  builder.AddCopy(source_path, dest_entry)
+def CopyDirContent(src, dest):
+  dest_entry = folder_map[dest]
+  for item in os.scandir(os.path.join(builder.sourcePath, src)):
+    if item.is_file():
+      builder.AddCopy(item.path, dest_entry)
+
+# Copy binaries.
+for task in HL2SDK.libs:
+  if task.target.platform == 'linux':
+    if task.target.arch == 'x86_64':
+      builder.AddCopy(task.binary, folder_map['linux64'])
+    else:
+      builder.AddCopy(task.binary, folder_map['linux'])
+  elif task.target.platform == 'windows':
+    if task.target.arch == 'x86_64':
+      builder.AddCopy(task.binary, folder_map['public/win64'])
+    else:
+      builder.AddCopy(task.binary, folder_map['public'])

configure.py

@@ -0,0 +1,11 @@
+import sys
+from ambuild2 import run
+
+parser = run.BuildParser(sourcePath = sys.path[0], api='2.2')
+parser.options.add_argument('--enable-debug', action='store_const', const='1', dest='debug',
+                        help='Enable debugging symbols')
+parser.options.add_argument('--enable-optimize', action='store_const', const='1', dest='opt',
+                        help='Enable optimization')
+parser.options.add_argument('--targets', type=str, dest='targets', default=None,
+		                help="Override the target architecture (use commas to separate multiple targets).")
+parser.Configure()

game/client/cdll_util.cpp

@@ -943,7 +943,7 @@ static unsigned char ComputeDistanceFade( C_BaseEntity *pEntity, float flMinDist
 
 	if( flMinDist > flMaxDist )
 	{
-		V_swap( flMinDist, flMaxDist );
+		::V_swap( flMinDist, flMaxDist );
 	}
 
 	// If a negative value is provided for the min fade distance, then base it off the max.

game/client/clientleafsystem.cpp

@@ -1722,8 +1722,8 @@ void CClientLeafSystem::SortEntities( const Vector &vecRenderOrigin, const Vecto
 		{
 			if( dists[i] > dists[i+stepSize] )
 			{
-				V_swap( pEntities[i], pEntities[i+stepSize] );
-				V_swap( dists[i], dists[i+stepSize] );
+				::V_swap( pEntities[i], pEntities[i+stepSize] );
+				::V_swap( dists[i], dists[i+stepSize] );
 
 				if( i == 0 )
 				{

game/client/clientshadowmgr.cpp

@@ -1138,7 +1138,7 @@ void CVisibleShadowList::PrioritySort()
 				flLargestArea = m_ShadowsInView[nIndex].m_flArea;
 			}
 		}
-		V_swap( m_PriorityIndex[i], m_PriorityIndex[nLargestInd] );
+		::V_swap( m_PriorityIndex[i], m_PriorityIndex[nLargestInd] );
 	}
 }
 

game/client/detailobjectsystem.cpp

@@ -1613,7 +1613,7 @@ void CDetailObjectSystem::UnserializeDetailSprites( CUtlBuffer& buf )
 		buf.Get( &m_DetailSpriteDict[i], sizeof(DetailSpriteDictLump_t) );
 		int flipi = m_DetailSpriteDictFlipped.AddToTail();
 		m_DetailSpriteDictFlipped[flipi] = m_DetailSpriteDict[i];
-		V_swap( m_DetailSpriteDictFlipped[flipi].m_TexUL.x, m_DetailSpriteDictFlipped[flipi].m_TexLR.x );
+		::V_swap( m_DetailSpriteDictFlipped[flipi].m_TexUL.x, m_DetailSpriteDictFlipped[flipi].m_TexLR.x );
 	}
 }
 

game/server/filters.cpp

@@ -536,7 +536,7 @@ bool CFilterEnemy::PassesProximityFilter( CBaseEntity *pCaller, CBaseEntity *pEn
 	float flSmallerRadius = m_flRadius;
 	if ( flSmallerRadius > flLargerRadius )
 	{
-		V_swap( flLargerRadius, flSmallerRadius );
+		::V_swap( flLargerRadius, flSmallerRadius );
 	}
 
 	float flDist;	

game/server/hltvdirector.cpp

@@ -562,7 +562,7 @@ void CHLTVDirector::CreateShotFromEvent( CGameEvent *event )
 			// if we show ineye view, show it more likely from killer
 			if ( RandomFloat(0,1) > (bInEye?0.3f:0.7f)  )
 			{
-				V_swap( attacker, victim );
+				::V_swap( attacker, victim );
 			}
 						
 			// hurting a victim is shown as chase more often

game/server/physconstraint.cpp

@@ -1280,7 +1280,7 @@ IPhysicsConstraint *CPhysSlideConstraint::CreateConstraint( IPhysicsConstraintGr
 		sliding.limitMax = DotProduct( axisDirection, m_axisEnd );
 		if ( sliding.limitMax < sliding.limitMin )
 		{
-			V_swap( sliding.limitMin, sliding.limitMax );
+			::V_swap( sliding.limitMin, sliding.limitMax );
 		}
 
 		// expand limits to make initial position of the attached object valid

game/server/props.cpp

@@ -4789,7 +4789,7 @@ void CPropDoorRotating::Spawn()
 	// that the model already be set.
 	if ( IsHingeOnLeft() )
 	{
-		V_swap( m_angRotationOpenForward, m_angRotationOpenBack );
+		::V_swap( m_angRotationOpenForward, m_angRotationOpenBack );
 	}
 
 	// Figure out our volumes of movement as this door opens

game/server/scratchpad_gamedll_helpers.cpp

@@ -15,7 +15,7 @@
 
 void ScratchPad_DrawWorldToScratchPad(
 	IScratchPad3D *pPad,
-	unsigned long flags )
+	uint32_t flags )
 {
 	pPad->SetRenderState( IScratchPad3D::RS_FillMode, IScratchPad3D::FillMode_Wireframe );
 
@@ -52,7 +52,7 @@ void ScratchPad_DrawWorldToScratchPad(
 
 void ScratchPad_DrawEntityToScratchPad(
 	IScratchPad3D *pPad,
-	unsigned long flags,
+	uint32_t flags,
 	CBaseEntity *pEnt,
 	const Vector &vColor )
 {

game/server/scratchpad_gamedll_helpers.h

@@ -27,12 +27,12 @@ class IScratchPad3D;
 // flags is a combination of the SPDRAWWORLD_ flags.
 void ScratchPad_DrawWorldToScratchPad(
 	IScratchPad3D *pPad,
-	unsigned long flags );
+	uint32_t flags );
 
 // Draw a specific entity into the scratch pad.
 void ScratchPad_DrawEntityToScratchPad(
 	IScratchPad3D *pPad,
-	unsigned long flags,
+	uint32_t flags,
 	CBaseEntity *pEnt,
 	const Vector &vColor );
 

game/shared/basecombatweapon_shared.h

@@ -86,7 +86,7 @@ class Color;
 
 namespace vgui2
 {
-	typedef unsigned long HFont;
+	typedef uint32_t HFont;
 }
 
 // -----------------------------------------

game/shared/choreoactor.cpp

@@ -1,4 +1,4 @@
-//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
+//========= Copyright <EFBFBD> 1996-2005, Valve Corporation, All rights reserved. ============//
 //
 // Purpose: 
 //
@@ -45,7 +45,7 @@ CChoreoActor& CChoreoActor::operator=( const CChoreoActor& src )
 	Q_strncpy( m_szName, src.m_szName, sizeof( m_szName ) );
 	Q_strncpy( m_szFacePoserModelName, src.m_szFacePoserModelName, sizeof( m_szFacePoserModelName ) );
 
-	for ( int i = 0; i < src.m_Channels.Size(); i++ )
+	for ( int i = 0; i < src.m_Channels.Count(); i++ )
 	{
 		CChoreoChannel *c = src.m_Channels[ i ];
 		CChoreoChannel *newChannel = new CChoreoChannel();
@@ -92,7 +92,7 @@ const char *CChoreoActor::GetName( void )
 //-----------------------------------------------------------------------------
 int CChoreoActor::GetNumChannels( void )
 {
-	return m_Channels.Size();
+	return m_Channels.Count();
 }
 
 //-----------------------------------------------------------------------------
@@ -102,7 +102,7 @@ int CChoreoActor::GetNumChannels( void )
 //-----------------------------------------------------------------------------
 CChoreoChannel *CChoreoActor::GetChannel( int channel )
 {
-	if ( channel < 0 || channel >= m_Channels.Size() )
+	if ( channel < 0 || channel >= m_Channels.Count() )
 	{
 		return NULL;
 	}
@@ -161,7 +161,7 @@ void CChoreoActor::SwapChannels( int c1, int c2 )
 //-----------------------------------------------------------------------------
 int CChoreoActor::FindChannelIndex( CChoreoChannel *channel )
 {
-	for ( int i = 0; i < m_Channels.Size(); i++ )
+	for ( int i = 0; i < m_Channels.Count(); i++ )
 	{
 		if ( channel == m_Channels[ i ] )
 		{

game/shared/choreochannel.cpp

@@ -1,4 +1,4 @@
-//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
+//========= Copyright <EFBFBD> 1996-2005, Valve Corporation, All rights reserved. ============//
 //
 // Purpose: 
 //
@@ -45,7 +45,7 @@ CChoreoChannel&	CChoreoChannel::operator=( const CChoreoChannel& src )
 {
 	m_bActive = src.m_bActive;
 	Q_strncpy( m_szName, src.m_szName, sizeof( m_szName ) );
-	for ( int i = 0; i < src.m_Events.Size(); i++ )
+	for ( int i = 0; i < src.m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = src.m_Events[ i ];
 		CChoreoEvent *newEvent = new CChoreoEvent( e->GetScene() );
@@ -83,7 +83,7 @@ const char *CChoreoChannel::GetName( void )
 //-----------------------------------------------------------------------------
 int CChoreoChannel::GetNumEvents( void )
 {
-	return m_Events.Size();
+	return m_Events.Count();
 }
 
 //-----------------------------------------------------------------------------
@@ -93,7 +93,7 @@ int CChoreoChannel::GetNumEvents( void )
 //-----------------------------------------------------------------------------
 CChoreoEvent *CChoreoChannel::GetEvent( int event )
 {
-	if ( event < 0 || event >= m_Events.Size() )
+	if ( event < 0 || event >= m_Events.Count() )
 	{
 		return NULL;
 	}
@@ -138,7 +138,7 @@ void CChoreoChannel::RemoveAllEvents()
 //-----------------------------------------------------------------------------
 int CChoreoChannel::FindEventIndex( CChoreoEvent *event )
 {
-	for ( int i = 0; i < m_Events.Size(); i++ )
+	for ( int i = 0; i < m_Events.Count(); i++ )
 	{
 		if ( event == m_Events[ i ] )
 		{

game/shared/choreoevent.cpp

@@ -1,4 +1,4 @@
-//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
+//========= Copyright <EFBFBD> 1996-2005, Valve Corporation, All rights reserved. ============//
 //
 // Purpose: 
 //
@@ -432,7 +432,7 @@ CFlexAnimationTrack::CFlexAnimationTrack( const CFlexAnimationTrack* src )
 	for ( int t = 0; t < 2; t++ )
 	{
 		m_Samples[ t ].Purge();
-		for ( int i = 0 ;i < src->m_Samples[ t ].Size(); i++ )
+		for ( int i = 0 ;i < src->m_Samples[ t ].Count(); i++ )
 		{
 			CExpressionSample s = src->m_Samples[ t ][ i ];
 			m_Samples[ t ].AddToTail( s );
@@ -527,7 +527,7 @@ int CFlexAnimationTrack::GetNumSamples( int type /*=0*/ )
 {
 	Assert( type == 0 || type == 1 );
 
-	return m_Samples[ type ].Size();
+	return m_Samples[ type ].Count();
 }
 
 //-----------------------------------------------------------------------------
@@ -931,9 +931,9 @@ void CFlexAnimationTrack::Resort( int type /*=0*/ )
 {
 	Assert( type == 0 || type == 1 );
 
-	for ( int i = 0; i < m_Samples[ type ].Size(); i++ )
+	for ( int i = 0; i < m_Samples[ type ].Count(); i++ )
 	{
-		for ( int j = i + 1; j < m_Samples[ type ].Size(); j++ )
+		for ( int j = i + 1; j < m_Samples[ type ].Count(); j++ )
 		{
 			CExpressionSample src = m_Samples[ type ][ i ];
 			CExpressionSample dest = m_Samples[ type ][ j ];
@@ -1116,7 +1116,7 @@ void CFlexAnimationTrack::RemoveOutOfRangeSamples( int type )
 	Assert( m_pEvent->HasEndTime() );
 	float duration = m_pEvent->GetDuration();
 
-	int c = m_Samples[ type ].Size();
+	int c = m_Samples[ type ].Count();
 	for ( int i = c-1; i >= 0; i-- )
 	{
 		CExpressionSample src = m_Samples[ type ][ i ];
@@ -1217,14 +1217,14 @@ CChoreoEvent& CChoreoEvent::operator=( const CChoreoEvent& src )
 	}
 
 	int i;
-	for ( i = 0; i < src.m_RelativeTags.Size(); i++ )
+	for ( i = 0; i < src.m_RelativeTags.Count(); i++ )
 	{	
 		CEventRelativeTag newtag( src.m_RelativeTags[ i ] );
 		newtag.SetOwner( this );
 		m_RelativeTags.AddToTail( newtag );
 	}
 
-	for ( i = 0; i < src.m_TimingTags.Size(); i++ )
+	for ( i = 0; i < src.m_TimingTags.Count(); i++ )
 	{	
 		CFlexTimingTag newtag( src.m_TimingTags[ i ] );
 		newtag.SetOwner( this );
@@ -1232,7 +1232,7 @@ CChoreoEvent& CChoreoEvent::operator=( const CChoreoEvent& src )
 	}
 	for ( t = 0; t < NUM_ABS_TAG_TYPES; t++ )
 	{
-		for ( i = 0; i < src.m_AbsoluteTags[ t ].Size(); i++ )
+		for ( i = 0; i < src.m_AbsoluteTags[ t ].Count(); i++ )
 		{
 			CEventAbsoluteTag newtag( src.m_AbsoluteTags[ t ][ i ] );
 			newtag.SetOwner( this );
@@ -1242,7 +1242,7 @@ CChoreoEvent& CChoreoEvent::operator=( const CChoreoEvent& src )
 
 	RemoveAllTracks();
 
-	for ( i = 0 ; i < src.m_FlexAnimationTracks.Size(); i++ )
+	for ( i = 0 ; i < src.m_FlexAnimationTracks.Count(); i++ )
 	{
 		CFlexAnimationTrack *newtrack = new CFlexAnimationTrack( src.m_FlexAnimationTracks[ i ] );
 		newtrack->SetEvent( this );
@@ -2382,7 +2382,7 @@ void CChoreoEvent::ClearAllRelativeTags( void )
 //-----------------------------------------------------------------------------
 int CChoreoEvent::GetNumRelativeTags( void )
 {
-	return m_RelativeTags.Size();
+	return m_RelativeTags.Count();
 }
 
 //-----------------------------------------------------------------------------
@@ -2392,7 +2392,7 @@ int CChoreoEvent::GetNumRelativeTags( void )
 //-----------------------------------------------------------------------------
 CEventRelativeTag *CChoreoEvent::GetRelativeTag( int tagnum )
 {
-	Assert( tagnum >= 0 && tagnum < m_RelativeTags.Size() );
+	Assert( tagnum >= 0 && tagnum < m_RelativeTags.Count() );
 	return &m_RelativeTags[ tagnum ];
 }
 
@@ -2413,7 +2413,7 @@ void CChoreoEvent::AddRelativeTag( const char *tagname, float percentage )
 //-----------------------------------------------------------------------------
 void CChoreoEvent::RemoveRelativeTag( const char *tagname )
 {
-	for ( int i = 0; i < m_RelativeTags.Size(); i++ )
+	for ( int i = 0; i < m_RelativeTags.Count(); i++ )
 	{
 		CEventRelativeTag *prt = &m_RelativeTags[ i ];
 		if ( !prt )
@@ -2434,7 +2434,7 @@ void CChoreoEvent::RemoveRelativeTag( const char *tagname )
 //-----------------------------------------------------------------------------
 CEventRelativeTag * CChoreoEvent::FindRelativeTag( const char *tagname )
 {
-	for ( int i = 0; i < m_RelativeTags.Size(); i++ )
+	for ( int i = 0; i < m_RelativeTags.Count(); i++ )
 	{
 		CEventRelativeTag *prt = &m_RelativeTags[ i ];
 		if ( !prt )
@@ -2517,7 +2517,7 @@ void CChoreoEvent::ClearAllTimingTags( void )
 //-----------------------------------------------------------------------------
 int CChoreoEvent::GetNumTimingTags( void )
 {
-	return m_TimingTags.Size();
+	return m_TimingTags.Count();
 }
 
 //-----------------------------------------------------------------------------
@@ -2527,7 +2527,7 @@ int CChoreoEvent::GetNumTimingTags( void )
 //-----------------------------------------------------------------------------
 CFlexTimingTag *CChoreoEvent::GetTimingTag( int tagnum )
 {
-	Assert( tagnum >= 0 && tagnum < m_TimingTags.Size() );
+	Assert( tagnum >= 0 && tagnum < m_TimingTags.Count() );
 	return &m_TimingTags[ tagnum ];
 }
 
@@ -2545,9 +2545,9 @@ void CChoreoEvent::AddTimingTag( const char *tagname, float percentage, bool loc
 	CFlexTimingTag temp( (CChoreoEvent *)0x1, "", 0.0f, false );
 
 	// ugly bubble sort
-	for ( int i = 0; i < m_TimingTags.Size(); i++ )
+	for ( int i = 0; i < m_TimingTags.Count(); i++ )
 	{
-		for ( int j = i + 1; j < m_TimingTags.Size(); j++ )
+		for ( int j = i + 1; j < m_TimingTags.Count(); j++ )
 		{
 			CFlexTimingTag *t1 = &m_TimingTags[ i ];
 			CFlexTimingTag *t2 = &m_TimingTags[ j ];
@@ -2568,7 +2568,7 @@ void CChoreoEvent::AddTimingTag( const char *tagname, float percentage, bool loc
 //-----------------------------------------------------------------------------
 void CChoreoEvent::RemoveTimingTag( const char *tagname )
 {
-	for ( int i = 0; i < m_TimingTags.Size(); i++ )
+	for ( int i = 0; i < m_TimingTags.Count(); i++ )
 	{
 		CFlexTimingTag *ptt = &m_TimingTags[ i ];
 		if ( !ptt )
@@ -2589,7 +2589,7 @@ void CChoreoEvent::RemoveTimingTag( const char *tagname )
 //-----------------------------------------------------------------------------
 CFlexTimingTag * CChoreoEvent::FindTimingTag( const char *tagname )
 {
-	for ( int i = 0; i < m_TimingTags.Size(); i++ )
+	for ( int i = 0; i < m_TimingTags.Count(); i++ )
 	{
 		CFlexTimingTag *ptt = &m_TimingTags[ i ];
 		if ( !ptt )
@@ -2626,7 +2626,7 @@ void CChoreoEvent::OnEndTimeChanged( void )
 //-----------------------------------------------------------------------------
 int CChoreoEvent::GetNumFlexAnimationTracks( void )
 {
-	return m_FlexAnimationTracks.Size();
+	return m_FlexAnimationTracks.Count();
 }
 
 //-----------------------------------------------------------------------------
@@ -2928,7 +2928,7 @@ void CChoreoEvent::ClearAllAbsoluteTags( AbsTagType type )
 //-----------------------------------------------------------------------------
 int CChoreoEvent::GetNumAbsoluteTags( AbsTagType type )
 {
-	return m_AbsoluteTags[ type ].Size();
+	return m_AbsoluteTags[ type ].Count();
 }
 
 //-----------------------------------------------------------------------------
@@ -2939,7 +2939,7 @@ int CChoreoEvent::GetNumAbsoluteTags( AbsTagType type )
 //-----------------------------------------------------------------------------
 CEventAbsoluteTag *CChoreoEvent::GetAbsoluteTag( AbsTagType type, int tagnum )
 {
-	Assert( tagnum >= 0 && tagnum < m_AbsoluteTags[ type ].Size() );
+	Assert( tagnum >= 0 && tagnum < m_AbsoluteTags[ type ].Count() );
 	return &m_AbsoluteTags[ type ][ tagnum ];
 }
 
@@ -2951,7 +2951,7 @@ CEventAbsoluteTag *CChoreoEvent::GetAbsoluteTag( AbsTagType type, int tagnum )
 //-----------------------------------------------------------------------------
 CEventAbsoluteTag *CChoreoEvent::FindAbsoluteTag( AbsTagType type, const char *tagname )
 {
-	for ( int i = 0; i < m_AbsoluteTags[ type ].Size(); i++ )
+	for ( int i = 0; i < m_AbsoluteTags[ type ].Count(); i++ )
 	{
 		CEventAbsoluteTag *ptag = &m_AbsoluteTags[ type ][ i ];
 		if ( !ptag )
@@ -2980,9 +2980,9 @@ void CChoreoEvent::AddAbsoluteTag( AbsTagType type, const char *tagname, float t
 	CEventAbsoluteTag temp( (CChoreoEvent *)0x1, "", 0.0f );
 
 	// ugly bubble sort
-	for ( int i = 0; i < m_AbsoluteTags[ type ].Size(); i++ )
+	for ( int i = 0; i < m_AbsoluteTags[ type ].Count(); i++ )
 	{
-		for ( int j = i + 1; j < m_AbsoluteTags[ type ].Size(); j++ )
+		for ( int j = i + 1; j < m_AbsoluteTags[ type ].Count(); j++ )
 		{
 			CEventAbsoluteTag *t1 = &m_AbsoluteTags[ type ][ i ];
 			CEventAbsoluteTag *t2 = &m_AbsoluteTags[ type ][ j ];
@@ -3004,7 +3004,7 @@ void CChoreoEvent::AddAbsoluteTag( AbsTagType type, const char *tagname, float t
 //-----------------------------------------------------------------------------
 void CChoreoEvent::RemoveAbsoluteTag( AbsTagType type, const char *tagname )
 {
-	for ( int i = 0; i < m_AbsoluteTags[ type ].Size(); i++ )
+	for ( int i = 0; i < m_AbsoluteTags[ type ].Count(); i++ )
 	{
 		CEventAbsoluteTag *ptag = &m_AbsoluteTags[ type ][ i ];
 		if ( !ptag )
@@ -3032,7 +3032,7 @@ bool CChoreoEvent::VerifyTagOrder( )
 	// Sort tags
 	CEventAbsoluteTag temp( (CChoreoEvent *)0x1, "", 0.0f );
 
-	for ( int i = 0; i < m_AbsoluteTags[ CChoreoEvent::ORIGINAL ].Size(); i++ )
+	for ( int i = 0; i < m_AbsoluteTags[ CChoreoEvent::ORIGINAL ].Count(); i++ )
 	{
 		CEventAbsoluteTag *ptag = &m_AbsoluteTags[ CChoreoEvent::ORIGINAL ][ i ];
 		if ( !ptag )
@@ -3044,7 +3044,7 @@ bool CChoreoEvent::VerifyTagOrder( )
 			continue;
 
 		bInOrder = false;
-		for ( int j = i + 1; j < m_AbsoluteTags[ CChoreoEvent::PLAYBACK ].Size(); j++ )
+		for ( int j = i + 1; j < m_AbsoluteTags[ CChoreoEvent::PLAYBACK ].Count(); j++ )
 		{
 			CEventAbsoluteTag *t2 = &m_AbsoluteTags[ CChoreoEvent::PLAYBACK ][ j ];
 
@@ -3529,9 +3529,9 @@ void CCurveData::Clear( void )
 //-----------------------------------------------------------------------------
 void CCurveData::Resort( ICurveDataAccessor *data )
 {
-	for ( int i = 0; i < m_Ramp.Size(); i++ )
+	for ( int i = 0; i < m_Ramp.Count(); i++ )
 	{
-		for ( int j = i + 1; j < m_Ramp.Size(); j++ )
+		for ( int j = i + 1; j < m_Ramp.Count(); j++ )
 		{
 			CExpressionSample src = m_Ramp[ i ];
 			CExpressionSample dest = m_Ramp[ j ];
@@ -3688,7 +3688,7 @@ bool CChoreoEvent::PreventTagOverlap( void )
 //-----------------------------------------------------------------------------
 CEventAbsoluteTag *CChoreoEvent::FindEntryTag( AbsTagType type )
 {
-	for ( int i = 0; i < m_AbsoluteTags[ type ].Size(); i++ )
+	for ( int i = 0; i < m_AbsoluteTags[ type ].Count(); i++ )
 	{
 		CEventAbsoluteTag *ptag = &m_AbsoluteTags[ type ][ i ];
 		if ( !ptag )
@@ -3709,7 +3709,7 @@ CEventAbsoluteTag *CChoreoEvent::FindEntryTag( AbsTagType type )
 //-----------------------------------------------------------------------------
 CEventAbsoluteTag *CChoreoEvent::FindExitTag( AbsTagType type )
 {
-	for ( int i = 0; i < m_AbsoluteTags[ type ].Size(); i++ )
+	for ( int i = 0; i < m_AbsoluteTags[ type ].Count(); i++ )
 	{
 		CEventAbsoluteTag *ptag = &m_AbsoluteTags[ type ][ i ];
 		if ( !ptag )

game/shared/choreoscene.cpp

@@ -1,4 +1,4 @@
-//===== Copyright © 1996-2005, Valve Corporation, All rights reserved. ======//
+//===== Copyright <EFBFBD> 1996-2005, Valve Corporation, All rights reserved. ======//
 //
 // Purpose: 
 //
@@ -102,7 +102,7 @@ void CChoreoScene::choreoprintf( int level, const char *fmt, ... )
 	}
 	else
 	{
-		printf( string );
+		printf( "%s", string );
 	}
 	
 	Msg( "%s", string );
@@ -128,7 +128,7 @@ CChoreoScene& CChoreoScene::operator=( const CChoreoScene& src )
 
 	// Delete existing
 	int i;
-	for ( i = 0; i < m_Actors.Size(); i++ )
+	for ( i = 0; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		Assert( a );
@@ -137,7 +137,7 @@ CChoreoScene& CChoreoScene::operator=( const CChoreoScene& src )
 
 	m_Actors.RemoveAll();
 
-	for ( i = 0; i < m_Events.Size(); i++ )
+	for ( i = 0; i < m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = m_Events[ i ];
 		Assert( e );
@@ -146,7 +146,7 @@ CChoreoScene& CChoreoScene::operator=( const CChoreoScene& src )
 
 	m_Events.RemoveAll();
 
-	for ( i = 0 ; i < m_Channels.Size(); i++ )
+	for ( i = 0 ; i < m_Channels.Count(); i++ )
 	{
 		CChoreoChannel *c = m_Channels[ i ];
 		Assert( c );
@@ -172,7 +172,7 @@ CChoreoScene& CChoreoScene::operator=( const CChoreoScene& src )
 	// Now copy the object tree
 	// First copy the global events
 
-	for ( i = 0; i < src.m_Events.Size(); i++ )
+	for ( i = 0; i < src.m_Events.Count(); i++ )
 	{
 		CChoreoEvent *event = src.m_Events[ i ];
 		if ( event->GetActor() == NULL )
@@ -186,7 +186,7 @@ CChoreoScene& CChoreoScene::operator=( const CChoreoScene& src )
 	}
 
 	// Finally, push actors, channels, events onto global stacks
-	for ( i = 0; i < src.m_Actors.Size(); i++ )
+	for ( i = 0; i < src.m_Actors.Count(); i++ )
 	{
 		CChoreoActor *actor = src.m_Actors[ i ];
 		CChoreoActor *newActor = AllocActor();
@@ -264,7 +264,7 @@ void CChoreoScene::Init( IChoreoEventCallback *callback )
 CChoreoScene::~CChoreoScene( void )
 {
 	int i;
-	for ( i = 0; i < m_Actors.Size(); i++ )
+	for ( i = 0; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		Assert( a );
@@ -273,7 +273,7 @@ CChoreoScene::~CChoreoScene( void )
 
 	m_Actors.RemoveAll();
 
-	for ( i = 0; i < m_Events.Size(); i++ )
+	for ( i = 0; i < m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = m_Events[ i ];
 		Assert( e );
@@ -282,7 +282,7 @@ CChoreoScene::~CChoreoScene( void )
 
 	m_Events.RemoveAll();
 
-	for ( i = 0 ; i < m_Channels.Size(); i++ )
+	for ( i = 0 ; i < m_Channels.Count(); i++ )
 	{
 		CChoreoChannel *c = m_Channels[ i ];
 		Assert( c );
@@ -390,7 +390,7 @@ void CChoreoScene::Print( void )
 	// Look for events that don't have actor/channel set
 	int i;
 
-	for ( i = 0 ; i < m_Events.Size(); i++ )
+	for ( i = 0 ; i < m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = m_Events[ i ];
 		if ( e->GetActor() )
@@ -399,7 +399,7 @@ void CChoreoScene::Print( void )
 		PrintEvent( 0, e );
 	}
 
-	for ( i = 0 ; i < m_Actors.Size(); i++ )
+	for ( i = 0 ; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )
@@ -478,7 +478,7 @@ CChoreoActor *CChoreoScene::AllocActor( void )
 //-----------------------------------------------------------------------------
 CChoreoActor *CChoreoScene::FindActor( const char *name )
 {
-	for ( int i = 0; i < m_Actors.Size(); i++ )
+	for ( int i = 0; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )
@@ -497,7 +497,7 @@ CChoreoActor *CChoreoScene::FindActor( const char *name )
 //-----------------------------------------------------------------------------
 int CChoreoScene::GetNumEvents( void )
 {
-	return m_Events.Size();
+	return m_Events.Count();
 }
 
 //-----------------------------------------------------------------------------
@@ -507,7 +507,7 @@ int CChoreoScene::GetNumEvents( void )
 //-----------------------------------------------------------------------------
 CChoreoEvent *CChoreoScene::GetEvent( int event )
 {
-	if ( event < 0 || event >= m_Events.Size() )
+	if ( event < 0 || event >= m_Events.Count() )
 		return NULL;
 
 	return m_Events[ event ];
@@ -519,7 +519,7 @@ CChoreoEvent *CChoreoScene::GetEvent( int event )
 //-----------------------------------------------------------------------------
 int CChoreoScene::GetNumActors( void )
 {
-	return m_Actors.Size();
+	return m_Actors.Count();
 }
 
 //-----------------------------------------------------------------------------
@@ -540,7 +540,7 @@ CChoreoActor *CChoreoScene::GetActor( int actor )
 //-----------------------------------------------------------------------------
 int CChoreoScene::GetNumChannels( void )
 {
-	return m_Channels.Size();
+	return m_Channels.Count();
 }
 
 //-----------------------------------------------------------------------------
@@ -619,9 +619,9 @@ void CCurveData::Parse( ISceneTokenProcessor *tokenizer, ICurveDataAccessor *dat
 			s->SetCurveType( curveType );
 		}
 
-		if ( samples.Size() >= 1 )
+		if ( samples.Count() >= 1 )
 		{
-			for ( int i = 0; i < samples.Size(); i++ )
+			for ( int i = 0; i < samples.Count(); i++ )
 			{
 				CExpressionSample sample = samples[ i ];
 
@@ -840,7 +840,7 @@ void CChoreoScene::ParseFlexAnimations( ISceneTokenProcessor *tokenizer, CChoreo
 			}
 		}
 		
-		if ( active || samples[ 0 ].Size() >= 1 )
+		if ( active || samples[ 0 ].Count() >= 1 )
 		{
 			// Add it in
 			CFlexAnimationTrack *track = e->AddTrack( flexcontroller );
@@ -853,7 +853,7 @@ void CChoreoScene::ParseFlexAnimations( ISceneTokenProcessor *tokenizer, CChoreo
 			
 			for ( int t = 0; t < ( combo ? 2 : 1 ); t++ )
 			{
-				for ( int i = 0; i < samples[ t ].Size(); i++ )
+				for ( int i = 0; i < samples[ t ].Count(); i++ )
 				{
 					CExpressionSample *sample = &samples[ t ][ i ];
 					
@@ -1527,7 +1527,7 @@ void CChoreoScene::InternalDetermineEventTypes()
 {
 	m_bitvecHasEventOfType.ClearAll();
 
-	for ( int i = 0 ; i < m_Actors.Size(); i++ )
+	for ( int i = 0 ; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )
@@ -1619,7 +1619,7 @@ void CChoreoScene::MarkForSaveAll( bool mark )
 	int i;
 
 	// Mark global events
-	for ( i = 0 ; i < m_Events.Size(); i++ )
+	for ( i = 0 ; i < m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = m_Events[ i ];
 		if ( e->GetActor() )
@@ -1629,7 +1629,7 @@ void CChoreoScene::MarkForSaveAll( bool mark )
 	}
 
 	// Recursively mark everything else
-	for ( i = 0 ; i < m_Actors.Size(); i++ )
+	for ( i = 0 ; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )
@@ -1652,7 +1652,7 @@ bool CChoreoScene::ExportMarkedToFile( const char *filename )
 
 	// Look for events that don't have actor/channel set
 	int i;
-	for ( i = 0 ; i < m_Events.Size(); i++ )
+	for ( i = 0 ; i < m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = m_Events[ i ];
 		if ( e->GetActor() )
@@ -1661,7 +1661,7 @@ bool CChoreoScene::ExportMarkedToFile( const char *filename )
 		FileSaveEvent( buf, 0, e );
 	}
 
-	for ( i = 0 ; i < m_Actors.Size(); i++ )
+	for ( i = 0 ; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )
@@ -1695,7 +1695,7 @@ bool CChoreoScene::SaveToFile( const char *filename )
 
 	// Look for events that don't have actor/channel set
 	int i;
-	for ( i = 0 ; i < m_Events.Size(); i++ )
+	for ( i = 0 ; i < m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = m_Events[ i ];
 		if ( e->GetActor() )
@@ -1704,7 +1704,7 @@ bool CChoreoScene::SaveToFile( const char *filename )
 		FileSaveEvent( buf, 0, e );
 	}
 
-	for ( i = 0 ; i < m_Actors.Size(); i++ )
+	for ( i = 0 ; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )
@@ -2219,7 +2219,7 @@ float CChoreoScene::FindAdjustedStartTime( void )
 
 	CChoreoEvent *e;
 
-	for ( int i = 0; i < m_Events.Size(); i++ )
+	for ( int i = 0; i < m_Events.Count(); i++ )
 	{
 		e = m_Events[ i ];
 
@@ -2251,7 +2251,7 @@ float CChoreoScene::FindAdjustedEndTime( void )
 
 	CChoreoEvent *e;
 
-	for ( int i = 0; i < m_Events.Size(); i++ )
+	for ( int i = 0; i < m_Events.Count(); i++ )
 	{
 		e = m_Events[ i ];
 
@@ -2290,7 +2290,7 @@ void CChoreoScene::ResetSimulation( bool forward /*= true*/, float starttime /*=
 	m_PauseEvents.RemoveAll();
 
 	// Put all items into the pending queue
-	for ( int i = 0; i < m_Events.Size(); i++ )
+	for ( int i = 0; i < m_Events.Count(); i++ )
 	{
 		e = m_Events[ i ];
 		e->ResetProcessing();
@@ -2317,7 +2317,7 @@ void CChoreoScene::ResetSimulation( bool forward /*= true*/, float starttime /*=
 	// choreoprintf( 0, "Start time %f\n", m_flCurrentTime );
 
 	m_flLastActiveTime = 0.0f;
-	m_nActiveEvents = m_Events.Size();
+	m_nActiveEvents = m_Events.Count();
 
 	m_flStartTime = starttime;
 	m_flEndTime = endtime;
@@ -2332,7 +2332,7 @@ bool CChoreoScene::CheckEventCompletion( void )
 
 	bool bAllCompleted = true;
 	// check all items in the active pending queue
-	for ( int i = 0; i < m_ActiveResumeConditions.Size(); i++ )
+	for ( int i = 0; i < m_ActiveResumeConditions.Count(); i++ )
 	{
 		e = m_ActiveResumeConditions[ i ];
 
@@ -2378,7 +2378,7 @@ CChoreoEvent *CChoreoScene::FindPauseBetweenTimes( float starttime, float endtim
 	CChoreoEvent *e;
 
 	// Iterate through all events in the scene
-	for ( int i = 0; i < m_PauseEvents.Size(); i++ )
+	for ( int i = 0; i < m_PauseEvents.Count(); i++ )
 	{
 		e = m_PauseEvents[ i ];
 		if ( !e )
@@ -2648,7 +2648,7 @@ void CChoreoScene::Think( float curtime )
 
 	// Iterate through all events in the scene
 	int i;
-	for ( i = 0; i < m_Events.Size(); i++ )
+	for ( i = 0; i < m_Events.Count(); i++ )
 	{
 		e = m_Events[ i ];
 		if ( !e )
@@ -2803,7 +2803,7 @@ void CChoreoScene::RemoveActor( CChoreoActor *actor )
 //-----------------------------------------------------------------------------
 int CChoreoScene::FindActorIndex( CChoreoActor *actor )
 {
-	for ( int i = 0; i < m_Actors.Size(); i++ )
+	for ( int i = 0; i < m_Actors.Count(); i++ )
 	{
 		if ( actor == m_Actors[ i ] )
 		{
@@ -2882,7 +2882,7 @@ void CChoreoScene::DeleteReferencedObjects( CChoreoEvent *event )
 //-----------------------------------------------------------------------------
 void CChoreoScene::DestroyActor( CChoreoActor *actor )
 {
-	int size = m_Actors.Size();
+	int size = m_Actors.Count();
 	for ( int i = size - 1; i >= 0; i-- )
 	{
 		CChoreoActor *a = m_Actors[ i ];
@@ -2901,7 +2901,7 @@ void CChoreoScene::DestroyActor( CChoreoActor *actor )
 //-----------------------------------------------------------------------------
 void CChoreoScene::DestroyChannel( CChoreoChannel *channel )
 {
-	int size = m_Channels.Size();
+	int size = m_Channels.Count();
 	for ( int i = size - 1; i >= 0; i-- )
 	{
 		CChoreoChannel *c = m_Channels[ i ];
@@ -2920,7 +2920,7 @@ void CChoreoScene::DestroyChannel( CChoreoChannel *channel )
 //-----------------------------------------------------------------------------
 void CChoreoScene::DestroyEvent( CChoreoEvent *event )
 {
-	int size = m_Events.Size();
+	int size = m_Events.Count();
 	for ( int i = size - 1; i >= 0; i-- )
 	{
 		CChoreoEvent *e = m_Events[ i ];
@@ -2998,7 +2998,7 @@ void CChoreoScene::GetSceneTimes( float& start, float& end )
 //-----------------------------------------------------------------------------
 void CChoreoScene::ReconcileTags( void )
 {
-	for ( int i = 0 ; i < m_Actors.Size(); i++ )
+	for ( int i = 0 ; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )
@@ -3055,7 +3055,7 @@ void CChoreoScene::ReconcileTags( void )
 //-----------------------------------------------------------------------------
 CChoreoEvent *CChoreoScene::FindTargetingEvent( const char *wavname, const char *name )
 {
-	for ( int i = 0 ; i < m_Actors.Size(); i++ )
+	for ( int i = 0 ; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )
@@ -3097,7 +3097,7 @@ CChoreoEvent *CChoreoScene::FindTargetingEvent( const char *wavname, const char
 //-----------------------------------------------------------------------------
 CEventRelativeTag *CChoreoScene::FindTagByName( const char *wavname, const char *name )
 {
-	for ( int i = 0 ; i < m_Actors.Size(); i++ )
+	for ( int i = 0 ; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )
@@ -3139,16 +3139,16 @@ CEventRelativeTag *CChoreoScene::FindTagByName( const char *wavname, const char
 //-----------------------------------------------------------------------------
 void CChoreoScene::ExportEvents( const char *filename, CUtlVector< CChoreoEvent * >& events )
 {
-	if ( events.Size() <= 0 )
+	if ( events.Count() <= 0 )
 		return;
 
 	// Create a serialization buffer
 	CUtlBuffer buf( 0, 0, CUtlBuffer::TEXT_BUFFER );
-	FilePrintf( buf, 0, "// Choreo version 1:  <%i> Exported Events\n", events.Size() );
+	FilePrintf( buf, 0, "// Choreo version 1:  <%i> Exported Events\n", events.Count() );
 
 	// Save out the selected events.
 	int i;
-	for ( i = 0 ; i < events.Size(); i++ )
+	for ( i = 0 ; i < events.Count(); i++ )
 	{
 		CChoreoEvent *e = events[ i ];
 		if ( !e->GetActor() )
@@ -3275,7 +3275,7 @@ float CChoreoScene::SnapTime( float t )
 //-----------------------------------------------------------------------------
 void CChoreoScene::ReconcileGestureTimes()
 {
-	for ( int i = 0 ; i < m_Actors.Size(); i++ )
+	for ( int i = 0 ; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )
@@ -3387,7 +3387,7 @@ bool CChoreoScene::Merge( CChoreoScene *other )
 
 	// Look for events that don't have actor/channel set
 	int i;
-	for ( i = 0 ; i < other->m_Events.Size(); i++ )
+	for ( i = 0 ; i < other->m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = other->m_Events[ i ];
 		if ( e->GetActor() )
@@ -3401,7 +3401,7 @@ bool CChoreoScene::Merge( CChoreoScene *other )
 		ecount++;
 	}
 
-	for ( i = 0 ; i < other->m_Actors.Size(); i++ )
+	for ( i = 0 ; i < other->m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = other->m_Actors[ i ];
 
@@ -3472,7 +3472,7 @@ bool CChoreoScene::Merge( CChoreoScene *other )
 //-----------------------------------------------------------------------------
 void CChoreoScene::ReconcileCloseCaption()
 {
-	for ( int i = 0 ; i < m_Actors.Size(); i++ )
+	for ( int i = 0 ; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )
@@ -3507,7 +3507,7 @@ void CChoreoScene::SetFileName( char const *fn )
 
 bool CChoreoScene::GetPlayingSoundName( char *pchBuff, int iBuffLength )
 {
-	for ( int i = 0; i < m_Events.Size(); i++ )
+	for ( int i = 0; i < m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = m_Events[ i ];
 		if ( e->GetType() == CChoreoEvent::SPEAK && e->IsProcessing() )
@@ -3525,7 +3525,7 @@ bool CChoreoScene::GetPlayingSoundName( char *pchBuff, int iBuffLength )
 //-----------------------------------------------------------------------------
 bool CChoreoScene::HasUnplayedSpeech()
 {
-	for ( int i = 0; i < m_Events.Size(); i++ )
+	for ( int i = 0; i < m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = m_Events[ i ];
 		if ( e->GetType() == CChoreoEvent::SPEAK )
@@ -3544,7 +3544,7 @@ bool CChoreoScene::HasUnplayedSpeech()
 //-----------------------------------------------------------------------------
 bool CChoreoScene::HasFlexAnimation()
 {
-	for ( int i = 0; i < m_Events.Size(); i++ )
+	for ( int i = 0; i < m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = m_Events[ i ];
 		if ( e->GetType() == CChoreoEvent::FLEXANIMATION )
@@ -3631,7 +3631,7 @@ void CChoreoScene::SaveToBinaryBuffer( CUtlBuffer& buf, unsigned int nTextVersio
 	// Look for events that don't have actor/channel set
 	CUtlVector< CChoreoEvent * > eventList;
 	int i;
-	for ( i = 0 ; i < m_Events.Size(); i++ )
+	for ( i = 0 ; i < m_Events.Count(); i++ )
 	{
 		CChoreoEvent *e = m_Events[ i ];
 		if ( e->GetActor() )
@@ -3651,7 +3651,7 @@ void CChoreoScene::SaveToBinaryBuffer( CUtlBuffer& buf, unsigned int nTextVersio
 
 	// Now serialize the actors themselves
 	CUtlVector< CChoreoActor * >	actorList;
-	for ( i = 0 ; i < m_Actors.Size(); i++ )
+	for ( i = 0 ; i < m_Actors.Count(); i++ )
 	{
 		CChoreoActor *a = m_Actors[ i ];
 		if ( !a )

game/shared/gamemovement.h

@@ -1,4 +1,4 @@
-//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
+//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
 //
 // Purpose: 
 //
@@ -42,9 +42,9 @@ public:
 	virtual void	StartTrackPredictionErrors( CBasePlayer *pPlayer );
 	virtual void	FinishTrackPredictionErrors( CBasePlayer *pPlayer );
 	virtual void	DiffPrint( char const *fmt, ... );
-	virtual const Vector&	GetPlayerMins( bool ducked ) const;
-	virtual const Vector&	GetPlayerMaxs( bool ducked ) const;
-	virtual const Vector&	GetPlayerViewOffset( bool ducked ) const;
+	virtual Vector  GetPlayerMins( bool ducked ) const;
+	virtual Vector	GetPlayerMaxs( bool ducked ) const;
+	virtual Vector	GetPlayerViewOffset( bool ducked ) const;
 
 // For sanity checking getting stuck on CMoveData::SetAbsOrigin
 	virtual void			TracePlayerBBox( const Vector& start, const Vector& end, unsigned int fMask, int collisionGroup, trace_t& pm );
@@ -105,8 +105,8 @@ protected:
 	// Implement this if you want to know when the player collides during OnPlayerMove
 	virtual void	OnTryPlayerMoveCollision( trace_t &tr ) {}
 
-	virtual const Vector&	GetPlayerMins( void ) const; // uses local player
-	virtual const Vector&	GetPlayerMaxs( void ) const; // uses local player
+	virtual Vector	GetPlayerMins( void ) const; // uses local player
+	virtual Vector	GetPlayerMaxs( void ) const; // uses local player
 
 	typedef enum
 	{

game/shared/gamerules.cpp

@@ -637,7 +637,7 @@ bool CGameRules::ShouldCollide( int collisionGroup0, int collisionGroup1 )
 	if ( collisionGroup0 > collisionGroup1 )
 	{
 		// swap so that lowest is always first
-		V_swap(collisionGroup0,collisionGroup1);
+		::V_swap(collisionGroup0,collisionGroup1);
 	}
 
 #ifndef HL2MP

game/shared/takedamageinfo.cpp

@@ -62,6 +62,8 @@ void CTakeDamageInfo::Init( CBaseEntity *pInflictor, CBaseEntity *pAttacker, CBa
 	m_flDamageBonus = 0.f;
 	m_bForceFriendlyFire = false;
 	m_flDamageForForce = 0.f;
+	
+	m_eCritType = kCritType_None;
 }
 
 CTakeDamageInfo::CTakeDamageInfo()
@@ -114,6 +116,18 @@ void CTakeDamageInfo::Set( CBaseEntity *pInflictor, CBaseEntity *pAttacker, CBas
 	Init( pInflictor, pAttacker, pWeapon, damageForce, damagePosition, vecReported, flDamage, bitsDamageType, iKillType );
 }
 
+void CTakeDamageInfo::SetCritType( ECritType type )
+{
+	if ( type == kCritType_None )
+	{
+		m_eCritType = kCritType_None;
+	}
+	else if ( type > m_eCritType )
+	{
+		m_eCritType = type;
+	}
+}
+
 //-----------------------------------------------------------------------------
 // Squirrel the damage value away as BaseDamage, which will later be used to 
 // calculate damage force. 

game/shared/takedamageinfo.h

@@ -98,6 +98,14 @@ public:
 	void			Set( CBaseEntity *pInflictor, CBaseEntity *pAttacker, const Vector &damageForce, const Vector &damagePosition, float flDamage, int bitsDamageType, int iKillType = 0, Vector *reportedPosition = NULL );
 	void			Set( CBaseEntity *pInflictor, CBaseEntity *pAttacker, CBaseEntity *pWeapon, const Vector &damageForce, const Vector &damagePosition, float flDamage, int bitsDamageType, int iKillType = 0, Vector *reportedPosition = NULL );
 	
+	enum ECritType
+	{
+		kCritType_None,
+		kCritType_MiniCrit,
+		kCritType_Crit,
+	};
+	void			SetCritType( ECritType type );
+
 	void			AdjustPlayerDamageInflictedForSkillLevel();
 	void			AdjustPlayerDamageTakenForSkillLevel();
 
@@ -133,6 +141,8 @@ protected:
 
 	float			m_flDamageForForce;
 	
+	ECritType		m_eCritType;
+
 	DECLARE_SIMPLE_DATADESC();
 };
 

game/shared/voice_gamemgr.cpp

@@ -162,7 +162,7 @@ bool CVoiceGameMgr::ClientCommand( CBasePlayer *pPlayer, const CCommand &args )
 	{
 		for(int i=1; i < args.ArgC(); i++)
 		{
-			unsigned long mask = 0;
+			uint32_t mask = 0;
 			sscanf( args[i], "%p", (void**)&mask);
 
 			if( i <= VOICE_MAX_PLAYERS_DW )

game/shared/voice_status.cpp

@@ -349,11 +349,11 @@ void CVoiceStatus::UpdateServerState(bool bForce)
 	Q_strncpy(str,"vban",sizeof(str));
 	bool bChange = false;
 
-	for(unsigned long dw=0; dw < VOICE_MAX_PLAYERS_DW; dw++)
+	for(uint32_t dw=0; dw < VOICE_MAX_PLAYERS_DW; dw++)
 	{	
-		unsigned long serverBanMask = 0;
-		unsigned long banMask = 0;
-		for(unsigned long i=0; i < 32; i++)
+		uint32_t serverBanMask = 0;
+		uint32_t banMask = 0;
+		for(uint32_t i=0; i < 32; i++)
 		{
 			int playerIndex = ( dw * 32 + i );
 			if ( playerIndex >= MAX_PLAYERS )
@@ -408,11 +408,11 @@ void CVoiceStatus::UpdateServerState(bool bForce)
 
 void CVoiceStatus::HandleVoiceMaskMsg(bf_read &msg)
 {
-	unsigned long dw;
+	uint32_t dw;
 	for(dw=0; dw < VOICE_MAX_PLAYERS_DW; dw++)
 	{
-		m_AudiblePlayers.SetDWord(dw, (unsigned long)msg.ReadLong());
-		m_ServerBannedPlayers.SetDWord(dw, (unsigned long)msg.ReadLong());
+		m_AudiblePlayers.SetDWord(dw, (uint32_t)msg.ReadLong());
+		m_ServerBannedPlayers.SetDWord(dw, (uint32_t)msg.ReadLong());
 
 		if( voice_clientdebug.GetInt())
 		{

linux_sdk/Makefile

@@ -1,175 +0,0 @@
-#
-# SDK Makefile for x86 Linux
-#
-#
-
-OS := $(shell uname -s)
-
-#############################################################################
-# Developer configurable items
-#############################################################################
-
-# the name of the mod binary (_i486.so is appended to the end)
-NAME = server
-
-# the location of the vcproj that builds the mod
-MOD_PROJ = ../game/server/server_scratch-2005.vcproj
-# the name of the mod configuration (typically <proj name>_<build type><build target>)
-MOD_CONFIG = Server\(SDK\)_ReleaseWin32
-
-# the directory the base binaries (tier0_i486.so, etc) are located
-# this should point to your orange box subfolder of where you have srcds installed.
-SRCDS_DIR = ~/srcds/orangebox
-
-# the path to your mods directory
-# set this so that 'make install' or 'make installrelease' will copy your binary over automatically.
-GAME_DIR = $(SRCDS_DIR)/scratchmod
-
-# compiler options (gcc 3.4.1 or above is required - 4.1.2+ recommended)
-ifeq "$(OS)" "Darwin"
-CC = /usr/bin/clang
-CPLUS = /usr/bin/clang++
-CLINK = /usr/bin/clang
-CPP_LIB =
-else
-CC = /usr/bin/gcc
-CPLUS = /usr/bin/g++
-CLINK = /usr/bin/gcc
-CPP_LIB = $(SRCDS_DIR)/bin/libstdc++.so.6 $(SRCDS_DIR)/bin/libgcc_s.so.1
-endif
-
-# put any compiler flags you want passed here
-USER_CFLAGS =
-
-# link flags for your mod, make sure to include any special libraries here
-LDFLAGS = "-lm -ldl $(LIB_DIR)/particles_i486.a $(LIB_DIR)/dmxloader_i486.a $(LIB_DIR)/mathlib_i486.a tier0_i486.so vstdlib_i486.so $(LIB_DIR)/tier1_i486.a $(LIB_DIR)/tier2_i486.a $(LIB_DIR)/tier3_i486.a $(LIB_DIR)/choreoobjects_i486.a steam_api_i486.so"
-
-# XERCES 2.6.0 or above ( http://xml.apache.org/xerces-c/ ) is used by the vcproj to makefile converter
-# it must be installed before being able to run this makefile
-# if you have xerces installed already you should be able to use the two lines below
-XERCES_INC_DIR = /usr/include
-XERCES_LIB_DIR = /usr/lib
-
-# Change this to true if you want to build debug binaries for everything
-# The only exception is the mod/game as MOD_CONFIG determines if it's a debug build or not
-DEBUG = false
-
-#############################################################################
-# Things below here shouldn't need to be altered
-#############################################################################
-MAKE = make
-AR = "ar rvs"
-
-# the dir we want to put binaries we build into
-BUILD_DIR = .
-# the place to put object files
-BUILD_OBJ_DIR = $(BUILD_DIR)/obj
-
-# the location of the source code
-SRC_DIR = ..
-
-# the location of the static libraries
-ifeq "$(OS)" "Darwin"
-LIB_DIR = $(SRC_DIR)/lib/mac
-else
-LIB_DIR = $(SRC_DIR)/lib/linux
-endif
-
-# the CPU target for the build, must be i486 for now
-ARCH = i486
-ARCH_CFLAGS = -mtune=i686 -march=pentium3 -mmmx -m32
-
-ifeq "$(OS)" "Darwin"
-DEFINES = -D_OSX -DOSX
-SHLIBEXT = dylib
-SHLIBLDFLAGS = -dynamiclib -mmacosx-version-min=10.5
-SHLIBSUFFIX = 
-else
-DEFINES = -D_LINUX -DLINUX
-SHLIBEXT = so
-SHLIBLDFLAGS = -shared -Wl,-Map,$@_map.txt -Wl
-SHLIBSUFFIX = _srv
-endif
-
-DEFINES +=-DVPROF_LEVEL=1 -DSWDS -D_finite=finite -Dstricmp=strcasecmp -D_stricmp=strcasecmp -D_strnicmp=strncasecmp \
-	-Dstrnicmp=strncasecmp -D_vsnprintf=vsnprintf -D_alloca=alloca -Dstrcmpi=strcasecmp
-UNDEF = -Usprintf -Ustrncpy -UPROTECTED_THINGS_ENABLE
-
-BASE_CFLAGS = -fno-strict-aliasing -Wall -Wsign-compare -Werror -Wno-conversion -Wno-overloaded-virtual -Wno-non-virtual-dtor -Wno-invalid-offsetof \
-	      -Wno-delete-non-virtual-dtor
-SHLIBCFLAGS = -fPIC
-
-# Flags passed to the c compiler
-CFLAGS = $(DEFINES) $(ARCH_CFLAGS) -O3 $(BASE_CFLAGS)
-ifdef USER_CFLAGS
-	CFLAGS += $(USER_CFLAGS)
-endif
-CFLAGS += $(UNDEF)
-
-# Debug flags
-DBG_DEFINES = "-D_DEBUG -DDEBUG"
-DBG_CFLAGS = "$(DEFINES) $(ARCH_CFLAGS) -g -ggdb $(BASE_CFLAGS) $(UNDEF)"
-
-# define list passed to make for the sub makefile
-BASE_DEFINES = CC=$(CC) AR=$(AR) CPLUS=$(CPLUS) CPP_LIB=$(CPP_LIB) DEBUG=$(DEBUG) \
-	BUILD_DIR=$(BUILD_DIR) BUILD_OBJ_DIR=$(BUILD_OBJ_DIR) SRC_DIR=$(SRC_DIR) \
-	LIB_DIR=$(LIB_DIR) SHLIBLDFLAGS="$(SHLIBLDFLAGS)" SHLIBEXT=$(SHLIBEXT) SHLIBSUFFIX=$(SHLIBSUFFIX) \
-	CLINK=$(CLINK) CFLAGS="$(CFLAGS)" DBG_CFLAGS=$(DBG_CFLAGS) LDFLAGS=$(LDFLAGS) \
-	DEFINES="$(DEFINES)" DBG_DEFINES=$(DBG_DEFINES) \
-	ARCH=$(ARCH) SRCDS_DIR=$(SRCDS_DIR) MOD_CONFIG=$(MOD_CONFIG) NAME=$(NAME) \
-	XERCES_INC_DIR=$(XERCES_INC_DIR) XERCES_LIB_DIR=$(XERCES_LIB_DIR)
-
-# Project Makefile
-MAKE_SERVER = Makefile.server
-MAKE_VCPM = Makefile.vcpm
-MAKE_PLUGIN = Makefile.plugin
-MAKE_TIER1 = Makefile.tier1
-MAKE_MATH = Makefile.mathlib
-MAKE_CHOREO = Makefile.choreo
-
-all: check vcpm mod
-
-check:
-	if [ -z "$(CC)" ]; then echo "Compiler not defined."; exit; fi
-	if [ ! -d $(BUILD_DIR) ];then mkdir -p $(BUILD_DIR);fi
-	cd $(BUILD_DIR)
-	if [ ! -e "$(LIB_DIR)/tier1_i486.a" ]; then $(MAKE) tier1;fi
-	if [ ! -e "$(LIB_DIR)/mathlib_i486.a" ]; then $(MAKE) mathlib;fi
-	if [ ! -e "$(LIB_DIR)/choreoobjects_i486.a" ]; then $(MAKE) choreo;fi
-	if [ ! -f "libtier0$(SHLIBSUFFIX).$(SHLIBEXT)" ]; then ln -fs $(LIB_DIR)/libtier0$(SHLIBSUFFIX).$(SHLIBEXT) .; fi
-	if [ ! -f "libvstdlib$(SHLIBSUFFIX).$(SHLIBEXT)" ]; then ln -fs $(LIB_DIR)/libvstdlib$(SHLIBSUFFIX).$(SHLIBEXT) .; fi
-	if [ ! -f "libsteam_api.$(SHLIBEXT)" ]; then ln -fs $(LIB_DIR)/libsteam_api.$(SHLIBEXT) .; fi
-
-vcpm: check
-	if [ ! -e "vcpm" ]; then $(MAKE) -f $(MAKE_VCPM) $(BASE_DEFINES);fi
-
-mod: check vcpm
-	./vcpm $(MOD_PROJ)
-	$(MAKE) -f $(MAKE_SERVER) $(BASE_DEFINES)
-
-plugin: check
-	$(MAKE) -f $(MAKE_PLUGIN) $(BASE_DEFINES)
-	
-tier1:
-	$(MAKE) -f $(MAKE_TIER1) $(BASE_DEFINES)
-
-mathlib:
-	$(MAKE) -f $(MAKE_MATH) $(BASE_DEFINES)
-
-choreo:
-	$(MAKE) -f $(MAKE_CHOREO) $(BASE_DEFINES)
-
-install:
-	cp -f $(NAME)_$(ARCH).$(SHLIBEXT) $(GAME_DIR)/bin/$(NAME)_$(ARCH).$(SHLIBEXT)
-
-installrelease:
-	cp -f $(NAME)_$(ARCH).$(SHLIBEXT) $(GAME_DIR)/bin/$(NAME)_$(ARCH).$(SHLIBEXT)
-	strip $(GAME_DIR)/bin/$(NAME)_$(ARCH).$(SHLIBEXT)
-
-clean:
-	$(MAKE) -f $(MAKE_VCPM) $(BASE_DEFINES) clean
-	$(MAKE) -f $(MAKE_PLUGIN) $(BASE_DEFINES) clean
-	$(MAKE) -f $(MAKE_SERVER) $(BASE_DEFINES) clean
-	$(MAKE) -f $(MAKE_TIER1) $(BASE_DEFINES) clean
-	$(MAKE) -f $(MAKE_MATH) $(BASE_DEFINES) clean
-	$(MAKE) -f $(MAKE_CHOREO) $(BASE_DEFINES) clean

linux_sdk/Makefile.choreo

@@ -1,57 +0,0 @@
-#
-# Choreoobjects Static Library Makefile
-#
-
-override NAME = choreoobjects
-
-LIB_SRC_DIR = $(SRC_DIR)/game/shared
-PUBLIC_SRC_DIR = $(SRC_DIR)/public
-TIER0_PUBLIC_SRC_DIR = $(SRC_DIR)/public/tier0
-TIER1_PUBLIC_SRC_DIR = $(SRC_DIR)/public/tier1
-UTIL_COMMON_SRC_DIR = $(SRC_DIR)/utils/common
-
-LIB_OBJ_DIR = $(BUILD_OBJ_DIR)/$(NAME)_$(ARCH)
-
-# Extension of linux static library
-override SHLIBEXT = a
-
-INCLUDEDIRS = -I$(LIB_SRC_DIR) -I$(PUBLIC_SRC_DIR) -I$(TIER0_PUBLIC_SRC_DIR) -I$(TIER1_PUBLIC_SRC_DIR) -I$(UTIL_COMMON_SRC_DIR) -D_LIB -DCHOREOOBJECTS_STATIC_LIB
-
-DO_CC = $(CPLUS) $(INCLUDEDIRS) -DARCH=$(ARCH)
-
-ifeq "$(DEBUG)" "true"
-	DO_CC += $(DBG_DEFINES) $(DBG_CFLAGS)
-else
-	DO_CC += -DNDEBUG $(CFLAGS)
-endif
-
-DO_CC += -o $@ -c $<
-
-#####################################################################
-
-LIB_OBJS= \
-	$(LIB_OBJ_DIR)/choreoactor.o \
-	$(LIB_OBJ_DIR)/choreochannel.o \
-	$(LIB_OBJ_DIR)/choreoevent.o \
-	$(LIB_OBJ_DIR)/choreoscene.o \
-	$(LIB_OBJ_DIR)/sceneimage.o \
-
-all: dirs $(NAME)_$(ARCH).$(SHLIBEXT)
-
-dirs:
-	-mkdir -p $(BUILD_OBJ_DIR)
-	-mkdir -p $(LIB_OBJ_DIR)
-
-$(NAME)_$(ARCH).$(SHLIBEXT): $(LIB_OBJS)
-	$(AR) $(LIB_DIR)/$@ $(LIB_OBJS)
-
-$(LIB_OBJ_DIR)/%.o: $(LIB_SRC_DIR)/%.cpp
-	$(DO_CC)
-
-install:
-	cp -f $(NAME)_$(ARCH).$(SHLIBEXT) $(LIB_DIR)/$(NAME)_$(ARCH).$(SHLIBEXT)
-
-clean:
-	-rm -rf $(LIB_OBJ_DIR)
-
-

linux_sdk/Makefile.mathlib

@@ -1,70 +0,0 @@
-#
-# Mathlin Static Library Makefile
-#
-
-override NAME = mathlib
-
-LIB_SRC_DIR = $(SRC_DIR)/$(NAME)
-PUBLIC_SRC_DIR = $(SRC_DIR)/public
-TIER0_PUBLIC_SRC_DIR = $(SRC_DIR)/public/tier0
-TIER1_PUBLIC_SRC_DIR = $(SRC_DIR)/public/tier1
-MATHLIB_PUBLIC_SRC_DIR = $(SRC_DIR)/public/mathlib
-
-LIB_OBJ_DIR = $(BUILD_OBJ_DIR)/$(NAME)_$(ARCH)
-
-# Extension of linux static library
-override SHLIBEXT = a
-
-INCLUDEDIRS = -I$(PUBLIC_SRC_DIR) -I$(TIER0_PUBLIC_SRC_DIR) -I$(TIER1_PUBLIC_SRC_DIR) -I$(MATHLIB_PUBLIC_SRC_DIR) -D_LIB
-
-DO_CC = $(CPLUS) $(INCLUDEDIRS) -DARCH=$(ARCH)
-
-ifeq "$(DEBUG)" "true"
-	DO_CC += $(DBG_DEFINES) $(DBG_CFLAGS)
-else
-	DO_CC += -DNDEBUG $(CFLAGS)
-endif
-
-DO_CC += -o $@ -c $<
-
-#####################################################################
-
-LIB_OBJS= \
-	$(LIB_OBJ_DIR)/anorms.o \
-	$(LIB_OBJ_DIR)/bumpvects.o \
-	$(LIB_OBJ_DIR)/color_conversion.o \
-	$(LIB_OBJ_DIR)/halton.o \
-	$(LIB_OBJ_DIR)/IceKey.o \
-	$(LIB_OBJ_DIR)/imagequant.o \
-	$(LIB_OBJ_DIR)/lightdesc.o \
-	$(LIB_OBJ_DIR)/mathlib_base.o \
-	$(LIB_OBJ_DIR)/polyhedron.o \
-	$(LIB_OBJ_DIR)/powsse.o \
-	$(LIB_OBJ_DIR)/quantize.o \
-	$(LIB_OBJ_DIR)/randsse.o \
-	$(LIB_OBJ_DIR)/simdvectormatrix.o \
-	$(LIB_OBJ_DIR)/sparse_convolution_noise.o \
-	$(LIB_OBJ_DIR)/sse.o \
-	$(LIB_OBJ_DIR)/sseconst.o \
-	$(LIB_OBJ_DIR)/ssenoise.o \
-	$(LIB_OBJ_DIR)/vector.o \
-	$(LIB_OBJ_DIR)/vmatrix.o \
-
-all: dirs $(NAME)_$(ARCH).$(SHLIBEXT)
-
-dirs:
-	-mkdir -p $(BUILD_OBJ_DIR)
-	-mkdir -p $(LIB_OBJ_DIR)
-
-$(NAME)_$(ARCH).$(SHLIBEXT): $(LIB_OBJS)
-	$(AR) $(LIB_DIR)/$@ $(LIB_OBJS)
-
-$(LIB_OBJ_DIR)/%.o: $(LIB_SRC_DIR)/%.cpp
-	$(DO_CC)
-
-install:
-	cp -f $(NAME)_$(ARCH).$(SHLIBEXT) $(LIB_DIR)/$(NAME)_$(ARCH).$(SHLIBEXT)
-
-clean:
-	-rm -rf $(LIB_OBJ_DIR)
-

linux_sdk/Makefile.plugin

@@ -1,57 +0,0 @@
-#
-# Sample server plugin for SRC engine
-#
-# October 2004, alfred@valvesoftware.com
-#
-
-override NAME = serverplugin_empty
-
-PLUGIN_SRC_DIR = $(SRC_DIR)/utils/serverplugin_sample
-PUBLIC_SRC_DIR = $(SRC_DIR)/public
-TIER0_PUBLIC_SRC_DIR = $(SRC_DIR)/public/tier0
-TIER1_PUBLIC_SRC_DIR = $(SRC_DIR)/public/tier1
-
-PLUGIN_OBJ_DIR = $(BUILD_OBJ_DIR)/serverplugin_empty_$(ARCH)
-TIER0_OBJ_DIR = $(PLUGIN_OBJ_DIR)/tier0
-
-INCLUDEDIRS = -I$(PUBLIC_SRC_DIR) -I$(TIER0_PUBLIC_SRC_DIR) -I$(TIER1_PUBLIC_SRC_DIR)
-LDFLAGS_PLG = -lm -ldl tier0_i486.so vstdlib_i486.so $(LIB_DIR)/mathlib_i486.a $(LIB_DIR)/tier1_i486.a $(LIB_DIR)/tier2_i486.a
-
-DO_CC = $(CPLUS) $(INCLUDEDIRS) -DARCH=$(ARCH)
-
-ifeq "$(DEBUG)" "true"
-	DO_CC += $(DBG_DEFINES) $(DBG_CFLAGS)
-else
-	DO_CC += -DNDEBUG $(CFLAGS)
-endif
-
-DO_CC += -o $@ -c $<
-
-#####################################################################
-
-PLUGIN_OBJS = \
-	$(PLUGIN_OBJ_DIR)/serverplugin_bot.o \
-	$(PLUGIN_OBJ_DIR)/serverplugin_empty.o \
-
-TIER0_OBJS = \
-	$(TIER0_OBJ_DIR)/memoverride.o \
-
-all: dirs $(NAME)_$(ARCH).$(SHLIBEXT)
-
-dirs:
-	-mkdir -p $(BUILD_OBJ_DIR)
-	-mkdir -p $(PLUGIN_OBJ_DIR)
-	-mkdir -p $(TIER0_OBJ_DIR)
-
-$(NAME)_$(ARCH).$(SHLIBEXT): $(PLUGIN_OBJS) $(TIER0_OBJS)
-	$(CLINK) -o $(BUILD_DIR)/$@ -m32 $(SHLIBLDFLAGS) $(PLUGIN_OBJS) $(TIER0_OBJS) $(PUBLIC_OBJS) $(CPP_LIB) $(LDFLAGS_PLG) $(CPP_LIB)
-
-$(PLUGIN_OBJ_DIR)/%.o: $(PLUGIN_SRC_DIR)/%.cpp
-	$(DO_CC)
-
-$(TIER0_OBJ_DIR)/%.o: $(TIER0_PUBLIC_SRC_DIR)/%.cpp
-	$(DO_CC)
-
-clean:
-	-rm -rf $(PLUGIN_OBJ_DIR)
-	-rm -f $(NAME)_$(ARCH).$(SHLIBEXT)

linux_sdk/Makefile.server

@@ -1,28 +0,0 @@
-#
-# wrapper Makefile for auto-generated make files
-#
-#
-
-#############################################################################
-# PROJECT MAKEFILES
-#############################################################################
-MAKE_FILE = Makefile.$(MOD_CONFIG)
--include $(MAKE_FILE)
-
-#############################################################################
-# The compiler command line for each src code file to compile
-#############################################################################
-DO_CC = $(CPLUS) $(INCLUDES) -DARCH=$(ARCH)
-
-ifeq (_DEBUG,$(findstring _DEBUG,$(CFLAGS)))
-	DO_CC += $(DEFINES) $(DBG_CFLAGS)
-else
-	DO_CC += $(CFLAGS)
-endif
-
-DO_CC += -o $@ -c $<
-
-clean:
-	rm -rf obj/$(NAME)_$(ARCH)
-	rm -f $(NAME)_$(ARCH).$(SHLIBEXT)
-

linux_sdk/Makefile.tier1

@@ -1,77 +0,0 @@
-#
-# Tier1 Static Library Makefile
-#
-
-override NAME = tier1
-
-LIB_SRC_DIR = $(SRC_DIR)/$(NAME)
-PUBLIC_SRC_DIR = $(SRC_DIR)/public
-TIER0_PUBLIC_SRC_DIR = $(SRC_DIR)/public/tier0
-TIER1_PUBLIC_SRC_DIR = $(SRC_DIR)/public/tier1
-
-LIB_OBJ_DIR=$(BUILD_OBJ_DIR)/$(NAME)_$(ARCH)
-
-# Extension of linux static library
-override SHLIBEXT = a
-
-INCLUDEDIRS = -I$(PUBLIC_SRC_DIR) -I$(TIER0_PUBLIC_SRC_DIR) -I$(TIER1_PUBLIC_SRC_DIR) -D_LIB -DTIER1_STATIC_LIB
-
-DO_CC = $(CPLUS) $(INCLUDEDIRS) -DARCH=$(ARCH)
-
-ifeq "$(DEBUG)" "true"
-	DO_CC += $(DBG_DEFINES) $(DBG_CFLAGS)
-else
-	DO_CC += -DNDEBUG $(CFLAGS)
-endif
-
-DO_CC += -o $@ -c $<
-
-#####################################################################
-
-LIB_OBJS= \
-	$(LIB_OBJ_DIR)/bitbuf.o \
-	$(LIB_OBJ_DIR)/byteswap.o \
-	$(LIB_OBJ_DIR)/characterset.o \
-	$(LIB_OBJ_DIR)/checksum_crc.o \
-	$(LIB_OBJ_DIR)/checksum_md5.o \
-	$(LIB_OBJ_DIR)/commandbuffer.o \
-	$(LIB_OBJ_DIR)/convar.o \
-	$(LIB_OBJ_DIR)/datamanager.o \
-	$(LIB_OBJ_DIR)/diff.o \
-	$(LIB_OBJ_DIR)/generichash.o \
-	$(LIB_OBJ_DIR)/interface.o \
-	$(LIB_OBJ_DIR)/KeyValues.o \
-	$(LIB_OBJ_DIR)/mempool.o \
-	$(LIB_OBJ_DIR)/NetAdr.o \
-	$(LIB_OBJ_DIR)/newbitbuf.o \
-	$(LIB_OBJ_DIR)/processor_detect.o \
-	$(LIB_OBJ_DIR)/rangecheckedvar.o \
-	$(LIB_OBJ_DIR)/stringpool.o \
-	$(LIB_OBJ_DIR)/strtools.o \
-	$(LIB_OBJ_DIR)/tier1.o \
-	$(LIB_OBJ_DIR)/tokenreader.o \
-	$(LIB_OBJ_DIR)/undiff.o \
-	$(LIB_OBJ_DIR)/uniqueid.o \
-	$(LIB_OBJ_DIR)/utlbuffer.o \
-	$(LIB_OBJ_DIR)/utlbufferutil.o \
-	$(LIB_OBJ_DIR)/utlstring.o \
-	$(LIB_OBJ_DIR)/utlsymbol.o \
-
-all: dirs $(NAME)_$(ARCH).$(SHLIBEXT)
-
-dirs:
-	-mkdir -p $(BUILD_OBJ_DIR)
-	-mkdir -p $(LIB_OBJ_DIR)
-
-$(NAME)_$(ARCH).$(SHLIBEXT): $(LIB_OBJS)
-	$(AR) $(LIB_DIR)/$@ $(LIB_OBJS)
-
-$(LIB_OBJ_DIR)/%.o: $(LIB_SRC_DIR)/%.cpp
-	$(DO_CC)
-
-install:
-	cp -f $(NAME)_$(ARCH).$(SHLIBEXT) $(LIB_DIR)/$(NAME)_$(ARCH).$(SHLIBEXT)
-
-clean:
-	-rm -rf $(LIB_OBJ_DIR)
-

linux_sdk/Makefile.vcpm

@@ -1,50 +0,0 @@
-#
-# VCProject file to Makefile converter
-#
-# November 2004, alfred@valvesoftware.com
-#
-
-VCPM_SRC_DIR = $(SRC_DIR)/utils/vprojtomake
-PUBLIC_SRC_DIR = $(SRC_DIR)/public
-TIER0_PUBLIC_SRC_DIR = $(SRC_DIR)/public/tier0
-TIER1_PUBLIC_SRC_DIR = $(SRC_DIR)/public/tier1
-UTIL_COMMON_SRC_DIR = $(SRC_DIR)/utils/common
-
-VCPM_OBJ_DIR = $(BUILD_OBJ_DIR)/vcpm
-
-INCLUDEDIRS = -I$(PUBLIC_SRC_DIR) -I$(XERCES_INC_DIR) -I$(TIER0_PUBLIC_SRC_DIR) -I$(TIER1_PUBLIC_SRC_DIR) -I$(UTIL_COMMON_SRC_DIR)
-LDFLAGS_VC = -lm -ldl -L$(XERCES_LIB_DIR) -lxerces-c tier0_i486.so vstdlib_i486.so $(LIB_DIR)/tier1_i486.a
-
-DO_CC = $(CPLUS) $(INCLUDEDIRS) -DARCH=$(ARCH)
-
-ifeq "$(DEBUG)" "true"
-	DO_CC += $(DBG_DEFINES) $(DBG_CFLAGS)
-else
-	DO_CC += -DNDEBUG $(CFLAGS)
-endif
-
-DO_CC += -o $@ -c $<
-
-#####################################################################
-
-VCPM_OBJS = \
-	$(VCPM_OBJ_DIR)/makefilecreator.o \
-	$(VCPM_OBJ_DIR)/vprojtomake.o \
-	$(VCPM_OBJ_DIR)/vcprojconvert.o \
-
-all: dirs vcpm
-
-dirs:
-	-mkdir -p $(BUILD_OBJ_DIR)
-	-mkdir -p $(VCPM_OBJ_DIR)
-
-vcpm: $(VCPM_OBJS)
-	$(CLINK) -m32 -o $(BUILD_DIR)/$@ $(VCPM_OBJS) $(CPP_LIB) $(LDFLAGS_VC)
-
-$(VCPM_OBJ_DIR)/%.o: $(VCPM_SRC_DIR)/%.cpp
-	$(DO_CC)
-
-clean:
-	-rm -rf $(VCPM_OBJ_DIR)
-	-rm -f vcpm
-

mathlib/3dnow.cpp

@@ -0,0 +1,197 @@
+//========= Copyright Valve Corporation, All rights reserved. ============//
+//
+// Purpose: 3DNow Math primitives.
+//
+//=====================================================================================//
+
+#include <math.h>
+#include <float.h>	// Needed for FLT_EPSILON
+#include "basetypes.h"
+#include <memory.h>
+#include "tier0/dbg.h"
+#include "mathlib/mathlib.h"
+#include "mathlib/amd3dx.h"
+#include "mathlib/vector.h"
+
+// memdbgon must be the last include file in a .cpp file!!!
+#include "tier0/memdbgon.h"
+
+#if !defined(COMPILER_MSVC64) && !defined(LINUX)
+// Implement for 64-bit Windows if needed.
+// Clang hits "fatal error: error in backend:" and other errors when trying
+// to compile the inline assembly below. 3DNow support is highly unlikely to
+// be useful/used, so it's not worth spending time on fixing.
+
+#pragma warning(disable:4244)   // "conversion from 'const int' to 'float', possible loss of data"
+#pragma warning(disable:4730)	// "mixing _m64 and floating point expressions may result in incorrect code"
+
+//-----------------------------------------------------------------------------
+// 3D Now Implementations of optimized routines:
+//-----------------------------------------------------------------------------
+float _3DNow_Sqrt(float x)
+{
+	Assert( s_bMathlibInitialized );
+	float	root = 0.f;
+#ifdef _WIN32
+	_asm
+	{
+		femms
+		movd		mm0, x
+		PFRSQRT		(mm1,mm0)
+		punpckldq	mm0, mm0
+		PFMUL		(mm0, mm1)
+		movd		root, mm0
+		femms
+	}
+#elif LINUX
+ 	__asm __volatile__( "femms" );
+ 	__asm __volatile__
+	(
+		"pfrsqrt    %y0, %y1 \n\t"
+		"punpckldq   %y1, %y1 \n\t"
+		"pfmul      %y1, %y0 \n\t"
+		: "=y" (root), "=y" (x)
+ 		:"0" (x)
+ 	);
+ 	__asm __volatile__( "femms" );
+#else
+#error
+#endif
+
+	return root;
+}
+
+// NJS FIXME: Need to test Recripricol squareroot performance and accuraccy
+// on AMD's before using the specialized instruction.
+float _3DNow_RSqrt(float x)
+{
+	Assert( s_bMathlibInitialized );
+
+	return 1.f / _3DNow_Sqrt(x);
+}
+
+
+float FASTCALL _3DNow_VectorNormalize (Vector& vec)
+{
+	Assert( s_bMathlibInitialized );
+	float *v = &vec[0];
+	float	radius = 0.f;
+
+	if ( v[0] || v[1] || v[2] )
+	{
+#ifdef _WIN32
+	_asm
+		{
+			mov			eax, v
+			femms
+			movq		mm0, QWORD PTR [eax]
+			movd		mm1, DWORD PTR [eax+8]
+			movq		mm2, mm0
+			movq		mm3, mm1
+			PFMUL		(mm0, mm0)
+			PFMUL		(mm1, mm1)
+			PFACC		(mm0, mm0)
+			PFADD		(mm1, mm0)
+			PFRSQRT		(mm0, mm1)
+			punpckldq	mm1, mm1
+			PFMUL		(mm1, mm0)
+			PFMUL		(mm2, mm0)
+			PFMUL		(mm3, mm0)
+			movq		QWORD PTR [eax], mm2
+			movd		DWORD PTR [eax+8], mm3
+			movd		radius, mm1
+			femms
+		}
+#elif LINUX	
+		long long a,c;
+    		int b,d;
+    		memcpy(&a,&vec[0],sizeof(a));
+    		memcpy(&b,&vec[2],sizeof(b));
+    		memcpy(&c,&vec[0],sizeof(c));
+    		memcpy(&d,&vec[2],sizeof(d));
+
+      		__asm __volatile__( "femms" );
+        	__asm __volatile__
+        	(
+        		"pfmul           %y3, %y3\n\t"
+        		"pfmul           %y0, %y0 \n\t"
+        		"pfacc           %y3, %y3 \n\t"
+        		"pfadd           %y3, %y0 \n\t"
+        		"pfrsqrt         %y0, %y3 \n\t"
+        		"punpckldq       %y0, %y0 \n\t"
+        		"pfmul           %y3, %y0 \n\t"
+        		"pfmul           %y3, %y2 \n\t"
+        		"pfmul           %y3, %y1 \n\t"
+        		: "=y" (radius), "=y" (c), "=y" (d)
+        		: "y" (a), "0" (b), "1" (c), "2" (d)
+        	);
+      		memcpy(&vec[0],&c,sizeof(c));
+      		memcpy(&vec[2],&d,sizeof(d));		
+        	__asm __volatile__( "femms" );
+
+#else
+#error
+#endif
+	}
+    return radius;
+}
+
+
+void FASTCALL _3DNow_VectorNormalizeFast (Vector& vec)
+{
+	_3DNow_VectorNormalize( vec );
+}
+
+
+// JAY: This complains with the latest processor pack
+#pragma warning(disable: 4730)
+
+float _3DNow_InvRSquared(const float* v)
+{
+	Assert( s_bMathlibInitialized );
+	float	r2 = 1.f;
+#ifdef _WIN32
+	_asm { // AMD 3DNow only routine
+		mov			eax, v
+		femms
+		movq		mm0, QWORD PTR [eax]
+		movd		mm1, DWORD PTR [eax+8]
+		movd		mm2, [r2]
+		PFMUL		(mm0, mm0)
+		PFMUL		(mm1, mm1)
+		PFACC		(mm0, mm0)
+		PFADD		(mm1, mm0)
+		PFMAX		(mm1, mm2)
+		PFRCP		(mm0, mm1)
+		movd		[r2], mm0
+		femms
+	}
+#elif LINUX
+		long long a,c;
+    		int b;
+    		memcpy(&a,&v[0],sizeof(a));
+    		memcpy(&b,&v[2],sizeof(b));
+    		memcpy(&c,&v[0],sizeof(c));
+
+      		__asm __volatile__( "femms" );
+        	__asm __volatile__
+        	(
+			"PFMUL          %y2, %y2 \n\t"
+                        "PFMUL          %y3, %y3 \n\t"
+                        "PFACC          %y2, %y2 \n\t"
+                        "PFADD          %y2, %y3 \n\t"
+                        "PFMAX          %y3, %y4 \n\t"
+                        "PFRCP          %y3, %y2 \n\t"
+                        "movq           %y2, %y0 \n\t"
+        		: "=y" (r2)
+        		: "0" (r2), "y" (a), "y" (b), "y" (c)
+        	);
+        	__asm __volatile__( "femms" );
+#else
+#error
+#endif
+
+	return r2;
+}
+
+#endif // COMPILER_MSVC64 

mathlib/3dnow.h

@@ -0,0 +1,16 @@
+//========= Copyright Valve Corporation, All rights reserved. ============//
+//
+// Purpose: 
+//
+//=====================================================================================//
+
+#ifndef _3DNOW_H
+#define _3DNOW_H
+
+float _3DNow_Sqrt(float x);
+float _3DNow_RSqrt(float x);
+float FASTCALL _3DNow_VectorNormalize (Vector& vec);
+void FASTCALL _3DNow_VectorNormalizeFast (Vector& vec);
+float _3DNow_InvRSquared(const float* v);
+
+#endif // _3DNOW_H

mathlib/AMBuilder

@@ -0,0 +1,48 @@
+# vim: set ts=2 sw=2 tw=99 noet ft=python:
+import os, sys
+
+def AddSourceFilesFromDir(path, files):
+  list = []
+  for file in files:
+    list.append(os.path.join(path, file))
+  return list
+
+builder.SetBuildFolder('/')
+
+project = builder.StaticLibraryProject('mathlib')
+project.sources = [
+	'3dnow.cpp',
+	'almostequal.cpp',
+	'anorms.cpp',
+	'bumpvects.cpp',
+	'color_conversion.cpp',
+	'halton.cpp',
+	'IceKey.cpp',
+	'imagequant.cpp',
+	'lightdesc.cpp',
+	'mathlib_base.cpp',
+	'polyhedron.cpp',
+	'powsse.cpp',
+	'quantize.cpp',
+	'randsse.cpp',
+	'simdvectormatrix.cpp',
+	'sparse_convolution_noise.cpp',
+	'spherical.cpp',
+	'sse.cpp',
+	'sseconst.cpp',
+	'ssenoise.cpp',
+	'vector.cpp',
+	'vmatrix.cpp',
+]
+
+for cxx in HL2SDK.targets:
+  binary = HL2SDK.ConfigureLibrary(project, cxx, builder)
+  compiler = binary.compiler
+
+  compiler.cxxincludes += [
+    os.path.join(builder.currentSourcePath, '..', 'public'),
+    os.path.join(builder.currentSourcePath, '..', 'public', 'tier0'),
+    os.path.join(builder.currentSourcePath, '..', 'public', 'mathlib')
+  ]
+
+HL2SDK.libs += builder.Add(project)

mathlib/IceKey.cpp

@@ -5,6 +5,9 @@
 #if !defined(_STATIC_LINKED) || defined(_SHARED_LIB)
 
 #include "mathlib/IceKey.H"
+#include <cstdint>
+
+#include "tier0/memdbgon.h"
 
 #ifdef _MSC_VER
 #pragma warning(disable: 4244)
@@ -13,12 +16,12 @@
 	/* Structure of a single round subkey */
 class IceSubkey {
     public:
-	unsigned long	val[3];
+	uint32_t	val[3];
 };
 
 
 	/* The S-boxes */
-static unsigned long	ice_sbox[4][1024];
+static uint32_t	ice_sbox[4][1024];
 static int		ice_sboxes_initialised = 0;
 
 
@@ -37,7 +40,7 @@ static const int	ice_sxor[4][4] = {
 				{0xea, 0xcb, 0x2e, 0x04}};
 
 	/* Permutation values for the P-box */
-static const unsigned long	ice_pbox[32] = {
+static const uint32_t	ice_pbox[32] = {
 		0x00000001, 0x00000080, 0x00000400, 0x00002000,
 		0x00080000, 0x00200000, 0x01000000, 0x40000000,
 		0x00000008, 0x00000020, 0x00000100, 0x00004000,
@@ -61,11 +64,11 @@ static const int	ice_keyrot[16] = {
 
 static unsigned int
 gf_mult (
-	register unsigned int	a,
-	register unsigned int	b,
-	register unsigned int	m
+	unsigned int	a,
+	unsigned int	b,
+	unsigned int	m
 ) {
-	register unsigned int	res = 0;
+	unsigned int	res = 0;
 
 	while (b) {
 	    if (b & 1)
@@ -87,12 +90,12 @@ gf_mult (
  * Raise the base to the power of 7, modulo m.
  */
 
-static unsigned long
+static uint32_t
 gf_exp7 (
-	register unsigned int	b,
+	unsigned int	b,
 	unsigned int		m
 ) {
-	register unsigned int	x;
+	unsigned int	x;
 
 	if (b == 0)
 	    return (0);
@@ -108,12 +111,12 @@ gf_exp7 (
  * Carry out the ICE 32-bit P-box permutation.
  */
 
-static unsigned long
+static uint32_t
 ice_perm32 (
-	register unsigned long	x
+	uint32_t	x
 ) {
-	register unsigned long		res = 0;
-	register const unsigned long	*pbox = ice_pbox;
+	uint32_t		res = 0;
+	const uint32_t	*pbox = ice_pbox;
 
 	while (x) {
 	    if (x & 1)
@@ -134,12 +137,12 @@ ice_perm32 (
 static void
 ice_sboxes_init (void)
 {
-	register int	i;
+	int	i;
 
 	for (i=0; i<1024; i++) {
 	    int			col = (i >> 1) & 0xff;
 	    int			row = (i & 0x1) | ((i & 0x200) >> 8);
-	    unsigned long	x;
+	    uint32_t	x;
 
 	    x = gf_exp7 (col ^ ice_sxor[0][row], ice_smod[0][row]) << 24;
 	    ice_sbox[0][i] = ice_perm32 (x);
@@ -201,13 +204,13 @@ IceKey::~IceKey ()
  * The single round ICE f function.
  */
 
-static unsigned long
+static uint32_t
 ice_f (
-	register unsigned long	p,
+	uint32_t	p,
 	const IceSubkey		*sk
 ) {
-	unsigned long	tl, tr;		/* Expanded 40-bit values */
-	unsigned long	al, ar;		/* Salted expanded 40-bit values */
+	uint32_t	tl, tr;		/* Expanded 40-bit values */
+	uint32_t	al, ar;		/* Salted expanded 40-bit values */
 
 					/* Left half expansion */
 	tl = ((p >> 16) & 0x3ff) | (((p >> 14) | (p << 18)) & 0xffc00);
@@ -241,15 +244,15 @@ IceKey::encrypt (
 	unsigned char		*ctext
 ) const
 {
-	register int		i;
-	register unsigned long	l, r;
+	int		i;
+	uint32_t	l, r;
 
-	l = (((unsigned long) ptext[0]) << 24)
-				| (((unsigned long) ptext[1]) << 16)
-				| (((unsigned long) ptext[2]) << 8) | ptext[3];
-	r = (((unsigned long) ptext[4]) << 24)
-				| (((unsigned long) ptext[5]) << 16)
-				| (((unsigned long) ptext[6]) << 8) | ptext[7];
+	l = (((uint32_t) ptext[0]) << 24)
+				| (((uint32_t) ptext[1]) << 16)
+				| (((uint32_t) ptext[2]) << 8) | ptext[3];
+	r = (((uint32_t) ptext[4]) << 24)
+				| (((uint32_t) ptext[5]) << 16)
+				| (((uint32_t) ptext[6]) << 8) | ptext[7];
 
 	for (i = 0; i < _rounds; i += 2) {
 	    l ^= ice_f (r, &_keysched[i]);
@@ -276,15 +279,15 @@ IceKey::decrypt (
 	unsigned char		*ptext
 ) const
 {
-	register int		i;
-	register unsigned long	l, r;
+	int		i;
+	uint32_t	l, r;
 
-	l = (((unsigned long) ctext[0]) << 24)
-				| (((unsigned long) ctext[1]) << 16)
-				| (((unsigned long) ctext[2]) << 8) | ctext[3];
-	r = (((unsigned long) ctext[4]) << 24)
-				| (((unsigned long) ctext[5]) << 16)
-				| (((unsigned long) ctext[6]) << 8) | ctext[7];
+	l = (((uint32_t) ctext[0]) << 24)
+				| (((uint32_t) ctext[1]) << 16)
+				| (((uint32_t) ctext[2]) << 8) | ctext[3];
+	r = (((uint32_t) ctext[4]) << 24)
+				| (((uint32_t) ctext[5]) << 16)
+				| (((uint32_t) ctext[6]) << 8) | ctext[7];
 
 	for (i = _rounds - 1; i > 0; i -= 2) {
 	    l ^= ice_f (r, &_keysched[i]);
@@ -314,20 +317,20 @@ IceKey::scheduleBuild (
 	int		i;
 
 	for (i=0; i<8; i++) {
-	    register int	j;
-	    register int	kr = keyrot[i];
+	    int	j;
+	    int	kr = keyrot[i];
 	    IceSubkey		*isk = &_keysched[n + i];
 
 	    for (j=0; j<3; j++)
 		isk->val[j] = 0;
 
 	    for (j=0; j<15; j++) {
-		register int	k;
-		unsigned long	*curr_sk = &isk->val[j % 3];
+		int	k;
+		uint32_t	*curr_sk = &isk->val[j % 3];
 
 		for (k=0; k<4; k++) {
 		    unsigned short	*curr_kb = &kb[(kr + k) & 3];
-		    register int	bit = *curr_kb & 1;
+		    int	bit = *curr_kb & 1;
 
 		    *curr_sk = (*curr_sk << 1) | bit;
 		    *curr_kb = (*curr_kb >> 1) | ((bit ^ 1) << 15);

mathlib/almostequal.cpp

@@ -0,0 +1,97 @@
+//========= Copyright Valve Corporation, All rights reserved. ============//
+//
+// Purpose: Fast ways to compare equality of two floats.  Assumes 
+// sizeof(float) == sizeof(int) and we are using IEEE format.
+//
+// Source:  http://www.cygnus-software.com/papers/comparingfloats/comparingfloats.htm
+//=====================================================================================//
+
+#include <float.h>
+#include <math.h>
+
+#include "mathlib/mathlib.h"
+
+static inline bool AE_IsInfinite(float a)
+{
+    const int kInfAsInt = 0x7F800000;
+
+    // An infinity has an exponent of 255 (shift left 23 positions) and
+    // a zero mantissa. There are two infinities - positive and negative.
+    if ((*(int*)&a & 0x7FFFFFFF) == kInfAsInt)
+        return true;
+    return false;
+}
+
+static inline bool AE_IsNan(float a)
+{
+    // a NAN has an exponent of 255 (shifted left 23 positions) and
+    // a non-zero mantissa.
+    int exp = *(int*)&a & 0x7F800000;
+    int mantissa = *(int*)&a & 0x007FFFFF;
+    if (exp == 0x7F800000 && mantissa != 0)
+        return true;
+    return false;
+}
+
+static inline int AE_Sign(float a)
+{
+    // The sign bit of a number is the high bit.
+    return (*(int*)&a) & 0x80000000;
+}
+
+// This is the 'final' version of the AlmostEqualUlps function.
+// The optional checks are included for completeness, but in many
+// cases they are not necessary, or even not desirable.
+bool AlmostEqual(float a, float b, int maxUlps)
+{
+    // There are several optional checks that you can do, depending
+    // on what behavior you want from your floating point comparisons.
+    // These checks should not be necessary and they are included
+    // mainly for completeness.
+
+    // If a or b are infinity (positive or negative) then
+    // only return true if they are exactly equal to each other -
+    // that is, if they are both infinities of the same sign.
+    // This check is only needed if you will be generating
+    // infinities and you don't want them 'close' to numbers
+    // near FLT_MAX.
+    if (AE_IsInfinite(a) || AE_IsInfinite(b))
+        return a == b;
+
+    // If a or b are a NAN, return false. NANs are equal to nothing,
+    // not even themselves.
+    // This check is only needed if you will be generating NANs
+    // and you use a maxUlps greater than 4 million or you want to
+    // ensure that a NAN does not equal itself.
+    if (AE_IsNan(a) || AE_IsNan(b))
+        return false;
+
+    // After adjusting floats so their representations are lexicographically
+    // ordered as twos-complement integers a very small positive number
+    // will compare as 'close' to a very small negative number. If this is
+    // not desireable, and if you are on a platform that supports
+    // subnormals (which is the only place the problem can show up) then
+    // you need this check.
+    // The check for a == b is because zero and negative zero have different
+    // signs but are equal to each other.
+    if (AE_Sign(a) != AE_Sign(b))
+        return a == b;
+
+    int aInt = *(int*)&a;
+    // Make aInt lexicographically ordered as a twos-complement int
+    if (aInt < 0)
+        aInt = 0x80000000 - aInt;
+    // Make bInt lexicographically ordered as a twos-complement int
+    int bInt = *(int*)&b;
+    if (bInt < 0)
+        bInt = 0x80000000 - bInt;
+
+    // Now we can compare aInt and bInt to find out how far apart a and b
+    // are.
+    int intDiff = abs(aInt - bInt);
+    if (intDiff <= maxUlps)
+        return true;
+    return false;
+}
+
+

mathlib/color_conversion.cpp

@@ -106,27 +106,23 @@ ALIGN128 float	power2_n[256] = 			// 2**(index - 128) / 255
 // You can use this to double check the exponent table and assert that 
 // the precomputation is correct.
 #ifdef DBGFLAG_ASSERT
-#ifdef _MSC_VER
+#ifdef _WIN32
 #pragma warning(push)
 #pragma warning( disable : 4189 ) // disable unused local variable warning
 #endif
-#ifdef __GNUC__
-__attribute__((unused)) static void CheckExponentTable()
-#else
 static void CheckExponentTable()
-#endif
 {
 	for( int i = 0; i < 256; i++ )
 	{
 		float testAgainst = pow( 2.0f, i - 128 ) / 255.0f;
 		float diff = testAgainst - power2_n[i] ;
 		float relativeDiff = diff / testAgainst;
-		Assert( sizeof(relativeDiff) > 0 && testAgainst == 0 ? 
+		Assert( testAgainst == 0 ? 
 				power2_n[i] < 1.16E-041 :
 				power2_n[i] == testAgainst );
 	}
 }
-#ifdef _MSC_VER
+#ifdef _WIN32
 #pragma warning(pop)
 #endif
 #endif
@@ -617,10 +613,10 @@ void VectorToColorRGBExp32( const Vector& vin, ColorRGBExp32 &c )
 		scalar = *reinterpret_cast<float *>(&fbits);
 	}
 
-	// we should never need to clamp:
-	Assert(vin.x * scalar <= 255.0f && 
-		   vin.y * scalar <= 255.0f && 
-		   vin.z * scalar <= 255.0f);
+	// We can totally wind up above 255 and that's okay--but above 256 would be right out.
+	Assert(vin.x * scalar < 256.0f && 
+		   vin.y * scalar < 256.0f && 
+		   vin.z * scalar < 256.0f);
 
 	// This awful construction is necessary to prevent VC2005 from using the 
 	// fldcw/fnstcw control words around every float-to-unsigned-char operation.

mathlib/imagequant.cpp

@@ -6,7 +6,7 @@
 //
 //=============================================================================//
 #include <quantize.h>
-#include <tier0/basetypes.h>
+#include <minmax.h>
 
 #define N_EXTRAVALUES 1
 #define N_DIMENSIONS (3+N_EXTRAVALUES)
@@ -46,7 +46,7 @@ void ColorQuantize(uint8 const *Image,
 					val1+=PIXEL(x,y,c)*ExtraValueXForms[i*3+c];
 				val1>>=8;
 				NthSample(s,y*Width+x,N_DIMENSIONS)->Value[c]=(uint8)
-					(MIN(255,MAX(0,val1)));
+					(V_min(255,V_max(0,val1)));
 			}
 		}
 	struct QuantizedValue *q=Quantize(s,Width*Height,N_DIMENSIONS,
@@ -76,7 +76,7 @@ void ColorQuantize(uint8 const *Image,
 					tryc+=Error[x][c][ErrorUse];
 					Error[x][c][ErrorUse]=0;
 				}
-				samp[c]=(uint8) MIN(255,MAX(0,tryc));
+				samp[c]=(uint8) V_min(255,V_max(0,tryc));
 			}
 			struct QuantizedValue *f=FindMatch(samp,3,Weights,q);
 			out_pixels[Width*y+x]=(uint8) (f->value);

mathlib/lightdesc.cpp

@@ -10,7 +10,7 @@
 
 void LightDesc_t::RecalculateDerivedValues(void)
 {
-	m_Flags=0;
+	m_Flags = LIGHTTYPE_OPTIMIZATIONFLAGS_DERIVED_VALUES_CALCED;
 	if (m_Attenuation0)
 		m_Flags|=LIGHTTYPE_OPTIMIZATIONFLAGS_HAS_ATTENUATION0;
 	if (m_Attenuation1)

mathlib/mathlib-2005.vcproj

@@ -1,399 +0,0 @@
-<?xml version="1.0" encoding="Windows-1252"?>
-<VisualStudioProject
-	ProjectType="Visual C++"
-	Version="8.00"
-	Name="mathlib"
-	ProjectGUID="{884C66F2-7F84-4570-AE6C-B634C1113D69}"
-	RootNamespace="mathlib"
-	>
-	<Platforms>
-		<Platform
-			Name="Win32"
-		/>
-	</Platforms>
-	<ToolFiles>
-	</ToolFiles>
-	<Configurations>
-		<Configuration
-			Name="Debug|Win32"
-			OutputDirectory=".\Debug"
-			IntermediateDirectory=".\Debug"
-			ConfigurationType="4"
-			CharacterSet="2"
-			>
-			<Tool
-				Name="VCPreBuildEventTool"
-				CommandLine=""
-				ExcludedFromBuild="false"
-			/>
-			<Tool
-				Name="VCCustomBuildTool"
-			/>
-			<Tool
-				Name="VCXMLDataGeneratorTool"
-			/>
-			<Tool
-				Name="VCWebServiceProxyGeneratorTool"
-			/>
-			<Tool
-				Name="VCMIDLTool"
-			/>
-			<Tool
-				Name="VCCLCompilerTool"
-				UseUnicodeResponseFiles="false"
-				Optimization="0"
-				AdditionalIncludeDirectories="..\public;..\public\tier0;..\public\tier1;..\public\mathlib"
-				PreprocessorDefinitions="WIN32;_WIN32;_DEBUG;DEBUG;_LIB;_CRT_SECURE_NO_DEPRECATE;_CRT_NONSTDC_NO_DEPRECATE"
-				StringPooling="true"
-				MinimalRebuild="true"
-				ExceptionHandling="0"
-				BasicRuntimeChecks="3"
-				RuntimeLibrary="1"
-				BufferSecurityCheck="false"
-				FloatingPointModel="2"
-				TreatWChar_tAsBuiltInType="true"
-				ForceConformanceInForLoopScope="true"
-				RuntimeTypeInfo="true"
-				OpenMP="false"
-				UsePrecompiledHeader="0"
-				ExpandAttributedSource="false"
-				AssemblerOutput="0"
-				AssemblerListingLocation="$(IntDir)/"
-				ObjectFile="$(IntDir)/"
-				ProgramDataBaseFileName="$(IntDir)/"
-				GenerateXMLDocumentationFiles="false"
-				BrowseInformation="0"
-				BrowseInformationFile="$(IntDir)/"
-				WarningLevel="4"
-				Detect64BitPortabilityProblems="true"
-				DebugInformationFormat="4"
-				CompileAs="2"
-				ErrorReporting="1"
-			/>
-			<Tool
-				Name="VCManagedResourceCompilerTool"
-			/>
-			<Tool
-				Name="VCResourceCompilerTool"
-			/>
-			<Tool
-				Name="VCPreLinkEventTool"
-				CommandLine=""
-				ExcludedFromBuild="false"
-			/>
-			<Tool
-				Name="VCLibrarianTool"
-				UseUnicodeResponseFiles="false"
-				OutputFile="..\lib\public\mathlib.lib"
-				SuppressStartupBanner="true"
-			/>
-			<Tool
-				Name="VCALinkTool"
-			/>
-			<Tool
-				Name="VCXDCMakeTool"
-				SuppressStartupBanner="true"
-			/>
-			<Tool
-				Name="VCBscMakeTool"
-				SuppressStartupBanner="true"
-				OutputFile="$(OutDir)/mathlib.bsc"
-			/>
-			<Tool
-				Name="VCFxCopTool"
-			/>
-			<Tool
-				Name="VCPostBuildEventTool"
-				ExcludedFromBuild="false"
-			/>
-		</Configuration>
-		<Configuration
-			Name="Release|Win32"
-			OutputDirectory=".\Release"
-			IntermediateDirectory=".\Release"
-			ConfigurationType="4"
-			CharacterSet="2"
-			>
-			<Tool
-				Name="VCPreBuildEventTool"
-				CommandLine=""
-				ExcludedFromBuild="false"
-			/>
-			<Tool
-				Name="VCCustomBuildTool"
-			/>
-			<Tool
-				Name="VCXMLDataGeneratorTool"
-			/>
-			<Tool
-				Name="VCWebServiceProxyGeneratorTool"
-			/>
-			<Tool
-				Name="VCMIDLTool"
-			/>
-			<Tool
-				Name="VCCLCompilerTool"
-				UseUnicodeResponseFiles="false"
-				Optimization="2"
-				InlineFunctionExpansion="2"
-				EnableIntrinsicFunctions="true"
-				FavorSizeOrSpeed="1"
-				AdditionalIncludeDirectories="..\public;..\public\tier0;..\public\tier1;..\public\mathlib"
-				PreprocessorDefinitions="WIN32;_WIN32;NDEBUG;_LIB;_CRT_SECURE_NO_DEPRECATE;_CRT_NONSTDC_NO_DEPRECATE"
-				StringPooling="true"
-				ExceptionHandling="0"
-				RuntimeLibrary="0"
-				BufferSecurityCheck="false"
-				EnableFunctionLevelLinking="true"
-				FloatingPointModel="2"
-				TreatWChar_tAsBuiltInType="true"
-				ForceConformanceInForLoopScope="true"
-				RuntimeTypeInfo="true"
-				OpenMP="false"
-				UsePrecompiledHeader="0"
-				ExpandAttributedSource="false"
-				AssemblerOutput="0"
-				AssemblerListingLocation="$(IntDir)/"
-				ObjectFile="$(IntDir)/"
-				ProgramDataBaseFileName="$(IntDir)/"
-				GenerateXMLDocumentationFiles="false"
-				BrowseInformation="0"
-				BrowseInformationFile="$(IntDir)/"
-				WarningLevel="4"
-				Detect64BitPortabilityProblems="true"
-				DebugInformationFormat="1"
-				CompileAs="2"
-				ErrorReporting="1"
-			/>
-			<Tool
-				Name="VCManagedResourceCompilerTool"
-			/>
-			<Tool
-				Name="VCResourceCompilerTool"
-			/>
-			<Tool
-				Name="VCPreLinkEventTool"
-				CommandLine=""
-				ExcludedFromBuild="false"
-			/>
-			<Tool
-				Name="VCLibrarianTool"
-				UseUnicodeResponseFiles="false"
-				OutputFile="..\lib\public\mathlib.lib"
-				SuppressStartupBanner="true"
-			/>
-			<Tool
-				Name="VCALinkTool"
-			/>
-			<Tool
-				Name="VCXDCMakeTool"
-				SuppressStartupBanner="true"
-			/>
-			<Tool
-				Name="VCBscMakeTool"
-				SuppressStartupBanner="true"
-				OutputFile="$(OutDir)/mathlib.bsc"
-			/>
-			<Tool
-				Name="VCFxCopTool"
-			/>
-			<Tool
-				Name="VCPostBuildEventTool"
-				ExcludedFromBuild="false"
-			/>
-		</Configuration>
-	</Configurations>
-	<References>
-	</References>
-	<Files>
-		<Filter
-			Name="Source Files"
-			Filter="cpp;c;cxx;rc;def;r;odl;idl;hpj;bat"
-			>
-			<File
-				RelativePath=".\anorms.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\bumpvects.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\color_conversion.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\halton.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\IceKey.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\imagequant.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\lightdesc.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\mathlib_base.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\polyhedron.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\powsse.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\quantize.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\randsse.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\simdvectormatrix.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\sparse_convolution_noise.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\sse.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\sseconst.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\ssenoise.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\vector.cpp"
-				>
-			</File>
-			<File
-				RelativePath=".\vmatrix.cpp"
-				>
-			</File>
-		</Filter>
-		<Filter
-			Name="Public Header Files"
-			Filter="h"
-			>
-			<File
-				RelativePath="..\public\mathlib\amd3dx.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\anorms.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\bumpvects.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\compressed_3d_unitvec.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\compressed_light_cube.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\compressed_vector.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\halton.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\IceKey.H"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\lightdesc.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\math_pfns.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\mathlib.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\noise.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\polyhedron.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\quantize.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\simdvectormatrix.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\ssemath.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\ssequaternion.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\vector.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\vector2d.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\vector4d.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\vmatrix.h"
-				>
-			</File>
-			<File
-				RelativePath="..\public\mathlib\vplane.h"
-				>
-			</File>
-		</Filter>
-		<Filter
-			Name="Header Files"
-			Filter="h;hpp;hxx;hm;inl"
-			>
-			<File
-				RelativePath=".\noisedata.h"
-				>
-			</File>
-			<File
-				RelativePath=".\sse.h"
-				>
-			</File>
-		</Filter>
-	</Files>
-	<Globals>
-	</Globals>
-</VisualStudioProject>

mathlib/mathlib_base.cpp

@@ -1,4 +1,4 @@
-//===== Copyright © 1996-2005, Valve Corporation, All rights reserved. ======//
+//===== Copyright <EFBFBD> 1996-2005, Valve Corporation, All rights reserved. ======//
 //
 // Purpose: Math primitives.
 //
@@ -6,6 +6,7 @@
 
 /// FIXME: As soon as all references to mathlib.c are gone, include it in here
 
+#include <algorithm>
 #include <math.h>
 #include <float.h>	// Needed for FLT_EPSILON
 
@@ -16,7 +17,7 @@
 #include "tier0/vprof.h"
 //#define _VPROF_MATHLIB
 
-#ifdef _MSC_VER
+#ifdef _WIN32
 #pragma warning(disable:4244)   // "conversion from 'const int' to 'float', possible loss of data"
 #pragma warning(disable:4730)	// "mixing _m64 and floating point expressions may result in incorrect code"
 #endif
@@ -24,6 +25,10 @@
 #include "mathlib/mathlib.h"
 #include "mathlib/vector.h"
 #if !defined( _X360 )
+#include "mathlib/amd3dx.h"
+#ifndef OSX
+#include "3dnow.h"
+#endif
 #include "sse.h"
 #endif
 
@@ -425,6 +430,33 @@ void MatrixSetColumn( const Vector &in, int column, matrix3x4_t& out )
 	out[2][column] = in.z;
 }
 
+void MatrixScaleBy ( const float flScale, matrix3x4_t &out )
+{
+	out[0][0] *= flScale;
+	out[1][0] *= flScale;
+	out[2][0] *= flScale;
+	out[0][1] *= flScale;
+	out[1][1] *= flScale;
+	out[2][1] *= flScale;
+	out[0][2] *= flScale;
+	out[1][2] *= flScale;
+	out[2][2] *= flScale;
+}
+
+void MatrixScaleByZero ( matrix3x4_t &out )
+{
+	out[0][0] = 0.0f;
+	out[1][0] = 0.0f;
+	out[2][0] = 0.0f;
+	out[0][1] = 0.0f;
+	out[1][1] = 0.0f;
+	out[2][1] = 0.0f;
+	out[0][2] = 0.0f;
+	out[1][2] = 0.0f;
+	out[2][2] = 0.0f;
+}
+
+
 
 int VectorCompare (const float *v1, const float *v2)
 {
@@ -564,53 +596,128 @@ void ConcatRotations (const float in1[3][3], const float in2[3][3], float out[3]
 				in1[2][2] * in2[2][2];
 }
 
+void ConcatTransforms_Aligned( const matrix3x4_t &m0, const matrix3x4_t &m1, matrix3x4_t &out )
+{
+	Assert( (((size_t)&m0) % 16) == 0 );
+	Assert( (((size_t)&m1) % 16) == 0 );
+	Assert( (((size_t)&out) % 16) == 0 );
+
+	fltx4 lastMask = *(fltx4 *)(&g_SIMD_ComponentMask[3]);
+	fltx4 rowA0 = LoadAlignedSIMD( m0.m_flMatVal[0] );
+	fltx4 rowA1 = LoadAlignedSIMD( m0.m_flMatVal[1] );
+	fltx4 rowA2 = LoadAlignedSIMD( m0.m_flMatVal[2] );
+
+	fltx4 rowB0 = LoadAlignedSIMD( m1.m_flMatVal[0] );
+	fltx4 rowB1 = LoadAlignedSIMD( m1.m_flMatVal[1] );
+	fltx4 rowB2 = LoadAlignedSIMD( m1.m_flMatVal[2] );
+
+	// now we have the rows of m0 and the columns of m1
+	// first output row
+	fltx4 A0 = SplatXSIMD(rowA0);
+	fltx4 A1 = SplatYSIMD(rowA0);
+	fltx4 A2 = SplatZSIMD(rowA0);
+	fltx4 mul00 = MulSIMD( A0, rowB0 );
+	fltx4 mul01 = MulSIMD( A1, rowB1 );
+	fltx4 mul02 = MulSIMD( A2, rowB2 );
+	fltx4 out0 = AddSIMD( mul00, AddSIMD(mul01,mul02) );
+
+	// second output row
+	A0 = SplatXSIMD(rowA1);
+	A1 = SplatYSIMD(rowA1);
+	A2 = SplatZSIMD(rowA1);
+	fltx4 mul10 = MulSIMD( A0, rowB0 );
+	fltx4 mul11 = MulSIMD( A1, rowB1 );
+	fltx4 mul12 = MulSIMD( A2, rowB2 );
+	fltx4 out1 = AddSIMD( mul10, AddSIMD(mul11,mul12) );
+
+	// third output row
+	A0 = SplatXSIMD(rowA2);
+	A1 = SplatYSIMD(rowA2);
+	A2 = SplatZSIMD(rowA2);
+	fltx4 mul20 = MulSIMD( A0, rowB0 );
+	fltx4 mul21 = MulSIMD( A1, rowB1 );
+	fltx4 mul22 = MulSIMD( A2, rowB2 );
+	fltx4 out2 = AddSIMD( mul20, AddSIMD(mul21,mul22) );
+
+	// add in translation vector
+	A0 = AndSIMD(rowA0,lastMask);
+	A1 = AndSIMD(rowA1,lastMask);
+	A2 = AndSIMD(rowA2,lastMask);
+	out0 = AddSIMD(out0, A0);
+	out1 = AddSIMD(out1, A1);
+	out2 = AddSIMD(out2, A2);
+
+	StoreAlignedSIMD( out.m_flMatVal[0], out0 );
+	StoreAlignedSIMD( out.m_flMatVal[1], out1 );
+	StoreAlignedSIMD( out.m_flMatVal[2], out2 );
+}
 
 /*
 ================
 R_ConcatTransforms
 ================
 */
+
 void ConcatTransforms (const matrix3x4_t& in1, const matrix3x4_t& in2, matrix3x4_t& out)
 {
-	Assert( s_bMathlibInitialized );
-	if ( &in1 == &out )
+#if 0
+	// test for ones that'll be 2x faster
+	if ( (((size_t)&in1) % 16) == 0 && (((size_t)&in2) % 16) == 0 && (((size_t)&out) % 16) == 0 )
 	{
-		matrix3x4_t in1b;
-		MatrixCopy( in1, in1b );
-		ConcatTransforms( in1b, in2, out );
+		ConcatTransforms_Aligned( in1, in2, out );
 		return;
 	}
-	if ( &in2 == &out )
-	{
-		matrix3x4_t in2b;
-		MatrixCopy( in2, in2b );
-		ConcatTransforms( in1, in2b, out );
-		return;
-	}
-	out[0][0] = in1[0][0] * in2[0][0] + in1[0][1] * in2[1][0] +
-				in1[0][2] * in2[2][0];
-	out[0][1] = in1[0][0] * in2[0][1] + in1[0][1] * in2[1][1] +
-				in1[0][2] * in2[2][1];
-	out[0][2] = in1[0][0] * in2[0][2] + in1[0][1] * in2[1][2] +
-				in1[0][2] * in2[2][2];
-	out[0][3] = in1[0][0] * in2[0][3] + in1[0][1] * in2[1][3] +
-				in1[0][2] * in2[2][3] + in1[0][3];
-	out[1][0] = in1[1][0] * in2[0][0] + in1[1][1] * in2[1][0] +
-				in1[1][2] * in2[2][0];
-	out[1][1] = in1[1][0] * in2[0][1] + in1[1][1] * in2[1][1] +
-				in1[1][2] * in2[2][1];
-	out[1][2] = in1[1][0] * in2[0][2] + in1[1][1] * in2[1][2] +
-				in1[1][2] * in2[2][2];
-	out[1][3] = in1[1][0] * in2[0][3] + in1[1][1] * in2[1][3] +
-				in1[1][2] * in2[2][3] + in1[1][3];
-	out[2][0] = in1[2][0] * in2[0][0] + in1[2][1] * in2[1][0] +
-				in1[2][2] * in2[2][0];
-	out[2][1] = in1[2][0] * in2[0][1] + in1[2][1] * in2[1][1] +
-				in1[2][2] * in2[2][1];
-	out[2][2] = in1[2][0] * in2[0][2] + in1[2][1] * in2[1][2] +
-				in1[2][2] * in2[2][2];
-	out[2][3] = in1[2][0] * in2[0][3] + in1[2][1] * in2[1][3] +
-				in1[2][2] * in2[2][3] + in1[2][3];
+#endif
+
+	fltx4 lastMask = *(fltx4 *)(&g_SIMD_ComponentMask[3]);
+	fltx4 rowA0 = LoadUnalignedSIMD( in1.m_flMatVal[0] );
+	fltx4 rowA1 = LoadUnalignedSIMD( in1.m_flMatVal[1] );
+	fltx4 rowA2 = LoadUnalignedSIMD( in1.m_flMatVal[2] );
+
+	fltx4 rowB0 = LoadUnalignedSIMD( in2.m_flMatVal[0] );
+	fltx4 rowB1 = LoadUnalignedSIMD( in2.m_flMatVal[1] );
+	fltx4 rowB2 = LoadUnalignedSIMD( in2.m_flMatVal[2] );
+
+	// now we have the rows of m0 and the columns of m1
+	// first output row
+	fltx4 A0 = SplatXSIMD(rowA0);
+	fltx4 A1 = SplatYSIMD(rowA0);
+	fltx4 A2 = SplatZSIMD(rowA0);
+	fltx4 mul00 = MulSIMD( A0, rowB0 );
+	fltx4 mul01 = MulSIMD( A1, rowB1 );
+	fltx4 mul02 = MulSIMD( A2, rowB2 );
+	fltx4 out0 = AddSIMD( mul00, AddSIMD(mul01,mul02) );
+
+	// second output row
+	A0 = SplatXSIMD(rowA1);
+	A1 = SplatYSIMD(rowA1);
+	A2 = SplatZSIMD(rowA1);
+	fltx4 mul10 = MulSIMD( A0, rowB0 );
+	fltx4 mul11 = MulSIMD( A1, rowB1 );
+	fltx4 mul12 = MulSIMD( A2, rowB2 );
+	fltx4 out1 = AddSIMD( mul10, AddSIMD(mul11,mul12) );
+
+	// third output row
+	A0 = SplatXSIMD(rowA2);
+	A1 = SplatYSIMD(rowA2);
+	A2 = SplatZSIMD(rowA2);
+	fltx4 mul20 = MulSIMD( A0, rowB0 );
+	fltx4 mul21 = MulSIMD( A1, rowB1 );
+	fltx4 mul22 = MulSIMD( A2, rowB2 );
+	fltx4 out2 = AddSIMD( mul20, AddSIMD(mul21,mul22) );
+
+	// add in translation vector
+	A0 = AndSIMD(rowA0,lastMask);
+	A1 = AndSIMD(rowA1,lastMask);
+	A2 = AndSIMD(rowA2,lastMask);
+	out0 = AddSIMD(out0, A0);
+	out1 = AddSIMD(out1, A1);
+	out2 = AddSIMD(out2, A2);
+
+	// write to output
+	StoreUnalignedSIMD( out.m_flMatVal[0], out0 );
+	StoreUnalignedSIMD( out.m_flMatVal[1], out1 );
+	StoreUnalignedSIMD( out.m_flMatVal[2], out2 );
 }
 
 
@@ -1357,7 +1464,9 @@ float Bias( float x, float biasAmt )
 	{
 		lastExponent = log( biasAmt ) * -1.4427f; // (-1.4427 = 1 / log(0.5))
 	}
-	return pow( x, lastExponent );
+	float fRet = pow( x, lastExponent );
+	Assert ( !IS_NAN( fRet ) );
+	return fRet;
 }
 
 
@@ -1373,7 +1482,9 @@ float Gain( float x, float biasAmt )
 
 float SmoothCurve( float x )
 {
-	return (1 - cos( x * M_PI )) * 0.5f;
+	// Actual smooth curve. Visualization:
+	// http://www.wolframalpha.com/input/?i=plot%5B+0.5+*+%281+-+cos%5B2+*+pi+*+x%5D%29+for+x+%3D+%280%2C+1%29+%5D
+	return 0.5f * (1 - cos( 2.0f * M_PI * x ) );
 }
 
 
@@ -1565,7 +1676,9 @@ float QuaternionAngleDiff( const Quaternion &p, const Quaternion &q )
 	QuaternionConjugate( q, qInv );
 	QuaternionMult( p, qInv, diff );
 
-	float sinang = sqrt( diff.x * diff.x + diff.y * diff.y + diff.z * diff.z );
+	// Note if the quaternion is slightly non-normalized the square root below may be more than 1,
+	// the value is clamped to one otherwise it may result in asin() returning an undefined result.
+	float sinang = MIN( 1.0f, sqrt( diff.x * diff.x + diff.y * diff.y + diff.z * diff.z ) );
 	float angle = RAD2DEG( 2 * asin( sinang ) );
 	return angle;
 #else
@@ -1665,7 +1778,7 @@ void QuaternionScale( const Quaternion &p, float t, Quaternion &q )
 	// FIXME: nick, this isn't overly sensitive to accuracy, and it may be faster to 
 	// use the cos part (w) of the quaternion (sin(omega)*N,cos(omega)) to figure the new scale.
 	float sinom = sqrt( DotProduct( &p.x, &p.x ) );
-	sinom = MIN( sinom, 1.f );
+	sinom = V_min( sinom, 1.f );
 
 	float sinsom = sin( asin( sinom ) * t );
 
@@ -1749,8 +1862,14 @@ void QuaternionMult( const Quaternion &p, const Quaternion &q, Quaternion &qt )
 
 
 void QuaternionMatrix( const Quaternion &q, const Vector &pos, matrix3x4_t& matrix )
+{
+#ifdef DBGFLAG_ASSERT
+	static bool s_bHushAsserts = !!CommandLine()->FindParm("-hushasserts");
+	if (!s_bHushAsserts)
 	{
 		Assert( pos.IsValid() );
+	}
+#endif
 
 	QuaternionMatrix( q, matrix );
 
@@ -1762,7 +1881,13 @@ void QuaternionMatrix( const Quaternion &q, const Vector &pos, matrix3x4_t& matr
 void QuaternionMatrix( const Quaternion &q, matrix3x4_t& matrix )
 {
 	Assert( s_bMathlibInitialized );
+#ifdef DBGFLAG_ASSERT
+	static bool s_bHushAsserts = !!CommandLine()->FindParm("-hushasserts");
+	if ( !s_bHushAsserts )
+	{
 		Assert( q.IsValid() );
+	}
+#endif
 
 #ifdef _VPROF_MATHLIB
 	VPROF_BUDGET( "QuaternionMatrix", "Mathlib" );
@@ -2949,9 +3074,9 @@ float CalcSqrDistanceToAABB( const Vector &mins, const Vector &maxs, const Vecto
 
 void CalcClosestPointOnAABB( const Vector &mins, const Vector &maxs, const Vector &point, Vector &closestOut )
 {
-	closestOut.x = clamp( point.x, mins.x, maxs.x );
-	closestOut.y = clamp( point.y, mins.y, maxs.y );
-	closestOut.z = clamp( point.z, mins.z, maxs.z );
+	closestOut.x = std::clamp( point.x, mins.x, maxs.x );
+	closestOut.y = std::clamp( point.y, mins.y, maxs.y );
+	closestOut.z = std::clamp( point.z, mins.z, maxs.z );
 }
 
 void CalcSqrDistAndClosestPointOnAABB( const Vector &mins, const Vector &maxs, const Vector &point, Vector &closestOut, float &distSqrOut )
@@ -3027,7 +3152,7 @@ void CalcClosestPointOnLineSegment( const Vector &P, const Vector &vLineA, const
 {
 	Vector vDir;
 	float t = CalcClosestPointToLineT( P, vLineA, vLineB, vDir );
-	t = clamp( t, 0, 1 );
+	t = std::clamp( t, 0.0f, 1.0f );
 	if ( outT ) 
 	{
 		*outT = t;
@@ -3099,7 +3224,7 @@ void CalcClosestPointOnLineSegment2D( const Vector2D &P, const Vector2D &vLineA,
 {
 	Vector2D vDir;
 	float t = CalcClosestPointToLineT2D( P, vLineA, vLineB, vDir );
-	t = clamp( t, 0, 1 );
+	t = std::clamp( t, 0.0f, 1.0f );
 	if ( outT )
 	{
 		*outT = t;
@@ -3210,7 +3335,7 @@ void MathLib_Init( float gamma, float texGamma, float brightness, int overbright
 
 #if !defined( _X360 )
 	// Grab the processor information:
-	const CPUInformation& pi = GetCPUInformation();
+	const CPUInformation& pi = *GetCPUInformation();
 
 	// Select the default generic routines.
 	pfSqrt = _sqrtf;
@@ -3233,12 +3358,33 @@ void MathLib_Init( float gamma, float texGamma, float brightness, int overbright
 		s_bMMXEnabled = false;
 	}
 
+	// SSE Generally performs better than 3DNow when present, so this is placed 
+	// first to allow SSE to override these settings.
+#if !defined( OSX ) && !defined( PLATFORM_WINDOWS_PC64 ) && !defined(LINUX)
+	if ( bAllow3DNow && pi.m_b3DNow )
+	{
+		s_b3DNowEnabled = true;
+
+		// Select the 3DNow specific routines if available;
+		pfVectorNormalize = _3DNow_VectorNormalize;
+		pfVectorNormalizeFast = _3DNow_VectorNormalizeFast;
+		pfInvRSquared = _3DNow_InvRSquared;
+		pfSqrt = _3DNow_Sqrt;
+		pfRSqrt = _3DNow_RSqrt;
+		pfRSqrtFast = _3DNow_RSqrt;
+	}
+	else
+#endif
+	{
 		s_b3DNowEnabled = false;
+	}
 
 	if ( bAllowSSE && pi.m_bSSE )
 	{
 		s_bSSEEnabled = true;
 
+#ifndef PLATFORM_WINDOWS_PC64
+		// These are not yet available.
 		// Select the SSE specific routines if available
 		pfVectorNormalize = _VectorNormalize;
 		pfVectorNormalizeFast = _SSE_VectorNormalizeFast;
@@ -3246,7 +3392,8 @@ void MathLib_Init( float gamma, float texGamma, float brightness, int overbright
 		pfSqrt = _SSE_Sqrt;
 		pfRSqrt = _SSE_RSqrtAccurate;
 		pfRSqrtFast = _SSE_RSqrtFast;
-#ifdef _WIN32
+#endif
+#ifdef PLATFORM_WINDOWS_PC32
 		pfFastSinCos = _SSE_SinCos;
 		pfFastCos = _SSE_cos;
 #endif
@@ -3259,7 +3406,7 @@ void MathLib_Init( float gamma, float texGamma, float brightness, int overbright
 	if ( bAllowSSE2 && pi.m_bSSE2 )
 	{
 		s_bSSE2Enabled = true;
-#ifdef _WIN32
+#ifdef PLATFORM_WINDOWS_PC32
 		pfFastSinCos = _SSE2_SinCos;
 		pfFastCos = _SSE2_cos;
 #endif
@@ -3268,7 +3415,7 @@ void MathLib_Init( float gamma, float texGamma, float brightness, int overbright
 	{
 		s_bSSE2Enabled = false;
 	}
-#endif
+#endif // !_X360
 
 	s_bMathlibInitialized = true;
 
@@ -3919,10 +4066,10 @@ void CalcTriangleTangentSpace( const Vector &p0, const Vector &p1, const Vector
 //-----------------------------------------------------------------------------
 void RGBtoHSV( const Vector &rgb, Vector &hsv )
 {
-	float flMax = MAX( rgb.x, rgb.y );
-	flMax = MAX( flMax, rgb.z );
-	float flMin = MIN( rgb.x, rgb.y );
-	flMin = MIN( flMin, rgb.z );
+	float flMax = V_max( rgb.x, rgb.y );
+	flMax = V_max( flMax, rgb.z );
+	float flMin = V_min( rgb.x, rgb.y );
+	flMin = V_min( flMin, rgb.z );
 
 	// hsv.z is the value
 	hsv.z = flMax;
@@ -4069,3 +4216,44 @@ void GetInterpolationData( float const *pKnotPositions,
 	*pInterpolationValue = FLerp( 0, 1, 0, flSizeOfGap, flOffsetFromStartOfGap );
 	return;
 }
+
+float RandomVectorInUnitSphere( Vector *pVector )
+{
+	// Guarantee uniform random distribution within a sphere
+	// Graphics gems III contains this algorithm ("Nonuniform random point sets via warping")
+	float u = ((float)rand() / VALVE_RAND_MAX);
+	float v = ((float)rand() / VALVE_RAND_MAX);
+	float w = ((float)rand() / VALVE_RAND_MAX);
+
+	float flPhi = acos( 1 - 2 * u );
+	float flTheta = 2 * M_PI * v;
+	float flRadius = powf( w, 1.0f / 3.0f );
+
+	float flSinPhi, flCosPhi;
+	float flSinTheta, flCosTheta;
+	SinCos( flPhi, &flSinPhi, &flCosPhi );
+	SinCos( flTheta, &flSinTheta, &flCosTheta );
+
+	pVector->x = flRadius * flSinPhi * flCosTheta;
+	pVector->y = flRadius * flSinPhi * flSinTheta;
+	pVector->z = flRadius * flCosPhi;
+	return flRadius;
+}
+
+float RandomVectorInUnitCircle( Vector2D *pVector )
+{
+	// Guarantee uniform random distribution within a sphere
+	// Graphics gems III contains this algorithm ("Nonuniform random point sets via warping")
+	float u = ((float)rand() / VALVE_RAND_MAX);
+	float v = ((float)rand() / VALVE_RAND_MAX);
+
+	float flTheta = 2 * M_PI * v;
+	float flRadius = powf( u, 1.0f / 2.0f );
+
+	float flSinTheta, flCosTheta;
+	SinCos( flTheta, &flSinTheta, &flCosTheta );
+
+	pVector->x = flRadius * flCosTheta;
+	pVector->y = flRadius * flSinTheta;
+	return flRadius;
+}

mathlib/polyhedron.cpp

@@ -34,7 +34,6 @@ CPolyhedron *ConvertLinkedGeometryToPolyhedron( GeneratePolyhedronFromPlanes_Uno
 //#define DEBUG_DUMP_POLYHEDRONS_TO_NUMBERED_GLVIEWS //dumps successfully generated polyhedrons
 
 #ifdef _DEBUG
-#include "filesystem.h"
 void DumpPolyhedronToGLView( const CPolyhedron *pPolyhedron, const char *pFilename, const VMatrix *pTransform );
 void DumpPlaneToGlView( const float *pPlane, float fGrayScale, const char *pszFileName, const VMatrix *pTransform );
 void DumpLineToGLView( const Vector &vPoint1, const Vector &vColor1, const Vector &vPoint2, const Vector &vColor2, float fThickness, FILE *pFile );
@@ -103,19 +102,19 @@ CPolyhedron_AllocByNew *CPolyhedron_AllocByNew::Allocate( unsigned short iVertic
 class CPolyhedron_TempMemory : public CPolyhedron
 {
 public:
-#ifdef _DEBUG
+#ifdef DBGFLAG_ASSERT
 	int iReferenceCount;
 #endif
 
 	virtual void Release( void )
 	{
-#ifdef _DEBUG
+#ifdef DBGFLAG_ASSERT
 		--iReferenceCount;
 #endif
 	}
 
 	CPolyhedron_TempMemory( void )
-#ifdef _DEBUG
+#ifdef DBGFLAG_ASSERT
 		: iReferenceCount( 0 )
 #endif
 	{ };
@@ -128,7 +127,7 @@ static CPolyhedron_TempMemory s_TempMemoryPolyhedron;
 CPolyhedron *GetTempPolyhedron( unsigned short iVertices, unsigned short iLines, unsigned short iIndices, unsigned short iPolygons ) //grab the temporary polyhedron. Avoids new/delete for quick work. Can only be in use by one chunk of code at a time
 {
 	AssertMsg( s_TempMemoryPolyhedron.iReferenceCount == 0, "Temporary polyhedron memory being rewritten before released" );
-#ifdef _DEBUG
+#ifdef DBGFLAG_ASSERT
 	++s_TempMemoryPolyhedron.iReferenceCount;
 #endif
 	s_TempMemoryPolyhedron_Buffer.SetCount( (sizeof( Vector ) * iVertices) +
@@ -857,8 +856,8 @@ const char * DumpPolyhedronCutHistory( const CUtlVector<CPolyhedron *> &DumpedHi
 
 #else
 
-#define AssertMsg_DumpPolyhedron(condition, message)
-#define Assert_DumpPolyhedron(condition)
+#define AssertMsg_DumpPolyhedron(condition, message) NULL;
+#define Assert_DumpPolyhedron(condition) NULL;
 
 #endif
 

mathlib/powsse.cpp

@@ -6,6 +6,10 @@
 
 #include "mathlib/ssemath.h"
 
+// NOTE: This has to be the last file included!
+#include "tier0/memdbgon.h"
+
+
 fltx4 Pow_FixedPoint_Exponent_SIMD( const fltx4 & x, int exponent)
 {
 	fltx4 rslt=Four_Ones;									// x^0=1.0
@@ -32,8 +36,61 @@ fltx4 Pow_FixedPoint_Exponent_SIMD( const fltx4 & x, int exponent)
 			break;
 	}
 	if (exponent<0)
-		return ReciprocalEstSIMD(rslt);							// pow(x,-b)=1/pow(x,b)
+		return ReciprocalEstSaturateSIMD(rslt);				// pow(x,-b)=1/pow(x,b)
 	else
 		return rslt;
 }
 
+
+
+
+/*
+ * (c) Ian Stephenson
+ *
+ * ian@dctsystems.co.uk
+ *
+ * Fast pow() reference implementation
+ */
+
+
+static float shift23=(1<<23);
+static float OOshift23=1.0/(1<<23);
+
+float FastLog2(float i)
+{
+	float LogBodge=0.346607f;
+	float x;
+	float y;
+	x=*(int *)&i;
+	x*= OOshift23; //1/pow(2,23);
+	x=x-127;
+
+	y=x-floorf(x);
+	y=(y-y*y)*LogBodge;
+	return x+y;
+}
+float FastPow2(float i)
+{
+	float PowBodge=0.33971f;
+	float x;
+	float y=i-floorf(i);
+	y=(y-y*y)*PowBodge;
+
+	x=i+127-y;
+	x*= shift23; //pow(2,23);
+	*(int*)&x=(int)x;
+	return x;
+}
+float FastPow(float a, float b)
+{
+	if (a <= OOshift23)
+	{
+		return 0.0f;
+	}
+	return FastPow2(b*FastLog2(a));
+}
+float FastPow10( float i )
+{
+	return FastPow2( i * 3.321928f );
+}
+

mathlib/quantize.cpp

@@ -18,11 +18,10 @@
 #endif
 
 #include <stdlib.h>
+#include <minmax.h>
 
 #include <math.h>
 
-#include <tier0/basetypes.h>
-
 static int current_ndims;
 static struct QuantizedValue *current_root;
 static int current_ssize;
@@ -412,8 +411,8 @@ static void Label(struct QuantizedValue *q, int updatecolor)
 		else
 			for(int i=0;i<current_ndims;i++)
 			{
-				q->Mins[i]=MIN(q->Children[0]->Mins[i],q->Children[1]->Mins[i]);
-				q->Maxs[i]=MAX(q->Children[0]->Maxs[i],q->Children[1]->Maxs[i]);
+				q->Mins[i]=V_min(q->Children[0]->Mins[i],q->Children[1]->Mins[i]);
+				q->Maxs[i]=V_max(q->Children[0]->Maxs[i],q->Children[1]->Maxs[i]);
 			}
 	}
 }    

mathlib/simdvectormatrix.cpp

@@ -48,7 +48,7 @@ void CSIMDVectorMatrix::CreateFromRGBA_FloatImageData(int srcwidth, int srcheigh
 			{
 				for(int cp=0;cp<4; cp++)
 				{
-					int real_cp=MIN( cp, ntrailing_pixels_per_source_line-1 );
+					int real_cp=V_min( cp, ntrailing_pixels_per_source_line-1 );
 					data_out[4*c+cp]= data_in[c+4*real_cp];
 				}
 			}

mathlib/spherical.cpp

@@ -0,0 +1,124 @@
+//========= Copyright Valve Corporation, All rights reserved. ============//
+//
+// Purpose: spherical math routines
+//
+//=====================================================================================//
+
+#include <math.h>
+#include <float.h>	// Needed for FLT_EPSILON
+#include "basetypes.h"
+#include <memory.h>
+#include "tier0/dbg.h"
+#include "mathlib/mathlib.h"
+#include "mathlib/vector.h"
+#include "mathlib/spherical_geometry.h"
+
+// memdbgon must be the last include file in a .cpp file!!!
+#include "tier0/memdbgon.h"
+
+float s_flFactorials[]={
+	1.,
+	1.,
+	2.,
+	6.,
+	24.,
+	120.,
+	720.,
+	5040.,
+	40320.,
+	362880.,
+	3628800.,
+	39916800.,
+	479001600.,
+	6227020800.,
+	87178291200.,
+	1307674368000.,
+	20922789888000.,
+	355687428096000.,
+	6402373705728000.,
+	121645100408832000.,
+	2432902008176640000.,
+	51090942171709440000.,
+	1124000727777607680000.,
+	25852016738884976640000.,
+	620448401733239439360000.,
+	15511210043330985984000000.,
+	403291461126605635584000000.,
+	10888869450418352160768000000.,
+	304888344611713860501504000000.,
+	8841761993739701954543616000000.,
+	265252859812191058636308480000000.,
+	8222838654177922817725562880000000.,
+	263130836933693530167218012160000000.,
+	8683317618811886495518194401280000000.
+};
+
+float AssociatedLegendrePolynomial( int nL, int nM, float flX )
+{
+	// evaluate associated legendre polynomial at flX, using recurrence relation
+	float flPmm = 1.;
+	if ( nM > 0 )
+	{
+		float flSomX2 = sqrt( ( 1 - flX ) * ( 1 + flX ) );
+		float flFact = 1.;
+		for( int i = 0 ; i < nM; i++ )
+		{
+			flPmm *= -flFact * flSomX2;
+			flFact += 2.0;
+		}
+	}
+	if ( nL == nM )
+		return flPmm;
+	float flPmmp1 = flX * ( 2.0 * nM + 1.0 ) * flPmm;
+	if ( nL == nM + 1 ) 
+		return flPmmp1;
+	float flPll = 0.;
+	for( int nLL = nM + 2 ; nLL <= nL; nLL++ )
+	{
+		flPll = ( ( 2.0 * nLL - 1.0 ) * flX * flPmmp1 - ( nLL + nM - 1.0 ) * flPmm ) * ( 1.0 / ( nLL - nM ) );
+		flPmm = flPmmp1;
+		flPmmp1 = flPll;
+	}
+	return flPll;
+}
+
+static float SHNormalizationFactor( int nL, int nM )
+{
+	double flTemp = ( ( 2. * nL + 1.0 ) * s_flFactorials[ nL - nM ] )/ ( 4. * M_PI * s_flFactorials[ nL + nM ] );
+	return sqrt( flTemp );
+}
+
+#define SQRT_2 1.414213562373095 
+
+FORCEINLINE float SphericalHarmonic( int nL, int nM, float flTheta, float flPhi, float flCosTheta )
+{
+	if ( nM == 0 )
+		return SHNormalizationFactor( nL, 0 ) * AssociatedLegendrePolynomial( nL, nM, flCosTheta );
+
+	if ( nM > 0 )
+		return SQRT_2 * SHNormalizationFactor( nL, nM ) * cos ( nM * flPhi ) *
+			AssociatedLegendrePolynomial( nL, nM, flCosTheta );
+
+	return 
+		SQRT_2 * SHNormalizationFactor( nL, -nM ) * sin( -nM * flPhi ) * AssociatedLegendrePolynomial( nL, -nM, flCosTheta );
+
+}
+
+float SphericalHarmonic( int nL, int nM, float flTheta, float flPhi )
+{
+	return SphericalHarmonic( nL, nM, flTheta, flPhi, cos( flTheta ) );
+}
+
+float SphericalHarmonic( int nL, int nM, Vector const &vecDirection )
+{
+	Assert( fabs( VectorLength( vecDirection ) - 1.0 ) < 0.0001 );
+	float flPhi = acos( vecDirection.z );
+	float flTheta = 0;
+	float S = Square( vecDirection.x ) + Square( vecDirection.y );
+	if ( S > 0 )
+	{
+		flTheta = atan2( vecDirection.y, vecDirection.x );
+	}
+	return SphericalHarmonic( nL, nM, flTheta, flPhi, cos( flTheta ) );
+}
+

mathlib/sse.cpp

@@ -1,4 +1,4 @@
-//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
+//========= Copyright <EFBFBD> 1996-2005, Valve Corporation, All rights reserved. ============//
 //
 // Purpose: SSE Math primitives.
 //
@@ -16,8 +16,13 @@
 // memdbgon must be the last include file in a .cpp file!!!
 #include "tier0/memdbgon.h"
 
+#ifndef COMPILER_MSVC64
+// Implement for 64-bit Windows if needed.
+
+#ifdef _WIN32
 static const uint32 _sincos_masks[]	  = { (uint32)0x0,  (uint32)~0x0 };
 static const uint32 _sincos_inv_masks[] = { (uint32)~0x0, (uint32)0x0 };
+#endif
 
 //-----------------------------------------------------------------------------
 // Macros and constants required by some of the SSE assembly:
@@ -35,20 +40,21 @@ static const uint32 _sincos_inv_masks[] = { (uint32)~0x0, (uint32)0x0 };
 
 	#define _PS_CONST(Name, Val) \
 		static const __declspec(align(16)) float _ps_##Name[4] = { Val, Val, Val, Val }
-#elif defined _LINUX || defined __APPLE__
+#elif POSIX
 	#define _PS_EXTERN_CONST(Name, Val) \
-		const __attribute__((aligned(16))) float _ps_##Name[4] = { Val, Val, Val, Val }
+		const float _ps_##Name[4] __attribute__((aligned(16))) = { Val, Val, Val, Val }
 
 	#define _PS_EXTERN_CONST_TYPE(Name, Type, Val) \
-		const __attribute__((aligned(16))) Type _ps_##Name[4] = { Val, Val, Val, Val }; \
+		const Type _ps_##Name[4]  __attribute__((aligned(16))) = { Val, Val, Val, Val }; \
 
 	#define _EPI32_CONST(Name, Val) \
-		static const __attribute__((aligned(16))) int32 _epi32_##Name[4] = { Val, Val, Val, Val }
+		static const int32 _epi32_##Name[4]  __attribute__((aligned(16))) = { Val, Val, Val, Val }
 
 	#define _PS_CONST(Name, Val) \
-		static const __attribute__((aligned(16))) float _ps_##Name[4] = { Val, Val, Val, Val }
+		static const float _ps_##Name[4]  __attribute__((aligned(16))) = { Val, Val, Val, Val }
 #endif
 
+#ifdef _WIN32
 _PS_EXTERN_CONST(am_0, 0.0f);
 _PS_EXTERN_CONST(am_1, 1.0f);
 _PS_EXTERN_CONST(am_m1, -1.0f);
@@ -59,8 +65,8 @@ _PS_EXTERN_CONST(am_pi_o_2, (float)(M_PI / 2.0));
 _PS_EXTERN_CONST(am_2_o_pi, (float)(2.0 / M_PI));
 _PS_EXTERN_CONST(am_pi_o_4, (float)(M_PI / 4.0));
 _PS_EXTERN_CONST(am_4_o_pi, (float)(4.0 / M_PI));
-_PS_EXTERN_CONST_TYPE(am_sign_mask, int32, 0x80000000);
-_PS_EXTERN_CONST_TYPE(am_inv_sign_mask, int32, ~0x80000000);
+_PS_EXTERN_CONST_TYPE(am_sign_mask, int32, static_cast<int32>(0x80000000));
+_PS_EXTERN_CONST_TYPE(am_inv_sign_mask, int32, static_cast<int32>(~0x80000000));
 _PS_EXTERN_CONST_TYPE(am_min_norm_pos,int32, 0x00800000);
 _PS_EXTERN_CONST_TYPE(am_mant_mask, int32, 0x7f800000);
 _PS_EXTERN_CONST_TYPE(am_inv_mant_mask, int32, ~0x7f800000);
@@ -72,6 +78,7 @@ _PS_CONST(sincos_p0, 0.15707963267948963959e1f);
 _PS_CONST(sincos_p1, -0.64596409750621907082e0f);
 _PS_CONST(sincos_p2, 0.7969262624561800806e-1f);
 _PS_CONST(sincos_p3, -0.468175413106023168e-2f);
+#endif
 
 #ifdef PFN_VECTORMA
 void  __cdecl _SSE_VectorMA( const float *start, float scale, const float *direction, float *dest );
@@ -90,14 +97,8 @@ float _SSE_Sqrt(float x)
 		sqrtss		xmm0, x
 		movss		root, xmm0
 	}
-#elif defined _LINUX || defined __APPLE__
-	__asm__ __volatile__(
-		"movss %1,%%xmm2\n"
-		"sqrtss %%xmm2,%%xmm1\n"
-		"movss %%xmm1,%0"
-       	: "=m" (root)
-		: "m" (x)
-	);
+#elif POSIX
+	_mm_store_ss( &root, _mm_sqrt_ss( _mm_load_ss( &x ) ) );
 #endif
 	return root;
 }
@@ -120,14 +121,21 @@ float _SSE_RSqrtAccurate(float x)
 	return (0.5f * rroot) * (3.f - (x * rroot) * rroot);
 }
 #else
+
+#ifdef POSIX
+const __m128  f3  = _mm_set_ss(3.0f);  // 3 as SSE value
+const __m128  f05 = _mm_set_ss(0.5f);  // 0.5 as SSE value
+#endif
+
 // Intel / Kipps SSE RSqrt.  Significantly faster than above.
 float _SSE_RSqrtAccurate(float a)
 {
+
+#ifdef _WIN32
 	float x;
 	float half = 0.5f;
 	float three = 3.f;
 
-#ifdef _WIN32
 	__asm
 	{
 		movss   xmm3, a;
@@ -143,26 +151,25 @@ float _SSE_RSqrtAccurate(float a)
 
 		movss   x,    xmm1;
 	}
-#elif defined _LINUX || defined __APPLE__
-	__asm__ __volatile__(
-		"movss   %1, %%xmm3 \n\t"
-        "movss   %2, %%xmm1 \n\t"
-        "movss   %3, %%xmm2 \n\t"
-        "rsqrtss %%xmm3, %%xmm0 \n\t"
-        "mulss   %%xmm0, %%xmm3 \n\t"
-        "mulss   %%xmm0, %%xmm1 \n\t"
-        "mulss   %%xmm0, %%xmm3 \n\t"
-        "subss   %%xmm3, %%xmm2 \n\t"
-        "mulss   %%xmm2, %%xmm1 \n\t"
-        "movss   %%xmm1, %0 \n\t"
-		: "=m" (x)
-		: "m" (a), "m" (half), "m" (three)
-);
+
+	return x;
+#elif POSIX	
+	__m128  xx = _mm_load_ss( &a );
+    __m128  xr = _mm_rsqrt_ss( xx );
+    __m128  xt;
+	
+    xt = _mm_mul_ss( xr, xr );
+    xt = _mm_mul_ss( xt, xx );
+    xt = _mm_sub_ss( f3, xt );
+    xt = _mm_mul_ss( xt, f05 );
+    xr = _mm_mul_ss( xr, xt );
+	
+    _mm_store_ss( &a, xr );
+    return a;
 #else
 	#error "Not Implemented"
 #endif
 
-	return x;
 }
 #endif
 
@@ -172,21 +179,15 @@ float _SSE_RSqrtFast(float x)
 {
 	Assert( s_bMathlibInitialized );
 
-	float rroot = 0.0f;
+	float rroot;
 #ifdef _WIN32
 	_asm
 	{
 		rsqrtss	xmm0, x
 		movss	rroot, xmm0
 	}
-#elif defined _LINUX || defined __APPLE__
-	 __asm__ __volatile__(
-		"rsqrtss %1, %%xmm0 \n\t"
-		"movss %%xmm0, %0 \n\t"
-		: "=m" (x)
-		: "m" (rroot)
-		: "%xmm0"
-	);
+#elif POSIX
+	__asm__ __volatile__( "rsqrtss %0, %1" : "=x" (rroot) : "x" (x) );
 #else
 #error
 #endif
@@ -202,11 +203,14 @@ float FASTCALL _SSE_VectorNormalize (Vector& vec)
 	// sice vec only has 3 floats, we can't "movaps" directly into it.
 #ifdef _WIN32
 	__declspec(align(16)) float result[4];
-#elif defined _LINUX || defined __APPLE__
-	__attribute__((aligned(16))) float result[4];
+#elif POSIX
+	 float result[4] __attribute__((aligned(16)));
 #endif
 
 	float *v = &vec[0];
+#ifdef _WIN32
+	float *r = &result[0];
+#endif
 
 	float	radius = 0.f;
 	// Blah, get rid of these comparisons ... in reality, if you have all 3 as zero, it shouldn't 
@@ -214,7 +218,6 @@ float FASTCALL _SSE_VectorNormalize (Vector& vec)
 	if ( v[0] || v[1] || v[2] )
 	{
 #ifdef _WIN32
-	float *r = &result[0];
 	_asm
 		{
 			mov			eax, v
@@ -239,7 +242,7 @@ float FASTCALL _SSE_VectorNormalize (Vector& vec)
 			mulps		xmm4, xmm1			// r4 = vx * 1/radius, vy * 1/radius, vz * 1/radius, X
 			movaps		[edx], xmm4			// v = vx * 1/radius, vy * 1/radius, vz * 1/radius, X
 		}
-#elif defined _LINUX || defined __APPLE__
+#elif POSIX
 		__asm__ __volatile__(
 #ifdef ALIGNED_VECTOR
             "movaps          %2, %%xmm4 \n\t"
@@ -262,6 +265,7 @@ float FASTCALL _SSE_VectorNormalize (Vector& vec)
             "movaps          %%xmm4, %1 \n\t"
             : "=m" (radius), "=m" (result)
             : "m" (*v)
+            : "xmm1", "xmm2", "xmm3", "xmm4"
  		);
 #else
 	#error "Not Implemented"
@@ -303,12 +307,13 @@ float _SSE_InvRSquared(const float* v)
 		shufps		xmm2, xmm2, 1		// x2 = vy * vy, X, X, X
 		addss		xmm1, xmm2			// x1 = (vx * vx) + (vy * vy), X, X, X
 		addss		xmm1, xmm3			// x1 = (vx * vx) + (vy * vy) + (vz * vz), X, X, X
-		maxss		xmm1, xmm5			// x1 = MAX( 1.0, x1 )
-		rcpss		xmm0, xmm1			// x0 = 1 / MAX( 1.0, x1 )
+		maxss		xmm1, xmm5			// x1 = max( 1.0, x1 )
+		rcpss		xmm0, xmm1			// x0 = 1 / max( 1.0, x1 )
 		movss		inv_r2, xmm0		// inv_r2 = x0
 	}
-#elif defined _LINUX || defined __APPLE__
+#elif POSIX
 		__asm__ __volatile__(
+		"movss			 %0, %%xmm5 \n\t"
 #ifdef ALIGNED_VECTOR
 		"movaps          %1, %%xmm4 \n\t"
 #else
@@ -324,8 +329,9 @@ float _SSE_InvRSquared(const float* v)
 		"maxss           %%xmm5, %%xmm1 \n\t"
         "rcpss           %%xmm1, %%xmm0 \n\t"
 		"movss           %%xmm0, %0 \n\t" 
-        : "=m" (inv_r2)
-        : "m" (*v), "m" (inv_r2)
+        : "+m" (inv_r2)
+        : "m" (*v)
+        : "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5"
  		);
 #else
 	#error "Not Implemented"
@@ -334,6 +340,50 @@ float _SSE_InvRSquared(const float* v)
 	return inv_r2;
 }
 
+
+#ifdef POSIX
+// #define _PS_CONST(Name, Val) static const ALIGN16 float _ps_##Name[4] ALIGN16_POST = { Val, Val, Val, Val }
+#define _PS_CONST_TYPE(Name, Type, Val) static const ALIGN16 Type _ps_##Name[4] ALIGN16_POST = { static_cast<Type>(Val), static_cast<Type>(Val), static_cast<Type>(Val), static_cast<Type>(Val) }
+
+_PS_CONST_TYPE(sign_mask, int, 0x80000000);
+_PS_CONST_TYPE(inv_sign_mask, int, ~0x80000000);
+
+
+#define _PI32_CONST(Name, Val)  static const ALIGN16 int _pi32_##Name[4]  ALIGN16_POST = { Val, Val, Val, Val }
+
+_PI32_CONST(1, 1);
+_PI32_CONST(inv1, ~1);
+_PI32_CONST(2, 2);
+_PI32_CONST(4, 4);
+#ifdef _WIN32
+_PI32_CONST(0x7f, 0x7f);
+#endif
+_PS_CONST(1  , 1.0f);
+_PS_CONST(0p5, 0.5f);
+
+_PS_CONST(minus_cephes_DP1, -0.78515625);
+_PS_CONST(minus_cephes_DP2, -2.4187564849853515625e-4);
+_PS_CONST(minus_cephes_DP3, -3.77489497744594108e-8);
+_PS_CONST(sincof_p0, -1.9515295891E-4);
+_PS_CONST(sincof_p1,  8.3321608736E-3);
+_PS_CONST(sincof_p2, -1.6666654611E-1);
+_PS_CONST(coscof_p0,  2.443315711809948E-005);
+_PS_CONST(coscof_p1, -1.388731625493765E-003);
+_PS_CONST(coscof_p2,  4.166664568298827E-002);
+_PS_CONST(cephes_FOPI, 1.27323954473516); // 4 / M_PI
+
+typedef union xmm_mm_union {
+	__m128 xmm;
+	__m64 mm[2];
+} xmm_mm_union;
+
+#define COPY_MM_TO_XMM(mm0_, mm1_, xmm_) { xmm_mm_union u; u.mm[0]=mm0_; u.mm[1]=mm1_; xmm_ = u.xmm; }
+
+typedef __m128 v4sf;  // vector of 4 float (sse1)
+typedef __m64 v2si;   // vector of 2 int (mmx)
+
+#endif
+
 void _SSE_SinCos(float x, float* s, float* c)
 {
 #ifdef _WIN32
@@ -421,8 +471,121 @@ void _SSE_SinCos(float x, float* s, float* c)
 		movss	[eax], xmm0
 		movss	[edx], xmm4
 	}
-#elif defined _LINUX || defined __APPLE__
-//	#warning "_SSE_sincos NOT implemented!"
+#elif POSIX
+	
+	Assert( "Needs testing, verify impl!\n" );
+	
+	v4sf  xx = _mm_load_ss( &x );
+	
+	v4sf xmm1, xmm2, xmm3 = _mm_setzero_ps(), sign_bit_sin, y;
+	v2si mm0, mm1, mm2, mm3, mm4, mm5;
+	sign_bit_sin = xx;
+	/* take the absolute value */
+	xx = _mm_and_ps(xx, *(v4sf*)_ps_inv_sign_mask);
+	/* extract the sign bit (upper one) */
+	sign_bit_sin = _mm_and_ps(sign_bit_sin, *(v4sf*)_ps_sign_mask);
+	
+	/* scale by 4/Pi */
+	y = _mm_mul_ps(xx, *(v4sf*)_ps_cephes_FOPI);
+	
+	/* store the integer part of y in mm2:mm3 */
+	xmm3 = _mm_movehl_ps(xmm3, y);
+	mm2 = _mm_cvttps_pi32(y);
+	mm3 = _mm_cvttps_pi32(xmm3);
+	
+	/* j=(j+1) & (~1) (see the cephes sources) */
+	mm2 = _mm_add_pi32(mm2, *(v2si*)_pi32_1);
+	mm3 = _mm_add_pi32(mm3, *(v2si*)_pi32_1);
+	mm2 = _mm_and_si64(mm2, *(v2si*)_pi32_inv1);
+	mm3 = _mm_and_si64(mm3, *(v2si*)_pi32_inv1);
+	
+	y = _mm_cvtpi32x2_ps(mm2, mm3);
+	
+	mm4 = mm2;
+	mm5 = mm3;
+	
+	/* get the swap sign flag for the sine */
+	mm0 = _mm_and_si64(mm2, *(v2si*)_pi32_4);
+	mm1 = _mm_and_si64(mm3, *(v2si*)_pi32_4);
+	mm0 = _mm_slli_pi32(mm0, 29);
+	mm1 = _mm_slli_pi32(mm1, 29);
+	v4sf swap_sign_bit_sin;
+	COPY_MM_TO_XMM(mm0, mm1, swap_sign_bit_sin);
+	
+	/* get the polynom selection mask for the sine */
+	
+	mm2 = _mm_and_si64(mm2, *(v2si*)_pi32_2);
+	mm3 = _mm_and_si64(mm3, *(v2si*)_pi32_2);
+	mm2 = _mm_cmpeq_pi32(mm2, _mm_setzero_si64());
+	mm3 = _mm_cmpeq_pi32(mm3, _mm_setzero_si64());
+	v4sf poly_mask;
+	COPY_MM_TO_XMM(mm2, mm3, poly_mask);
+	
+	/* The magic pass: "Extended precision modular arithmetic" 
+	 x = ((x - y * DP1) - y * DP2) - y * DP3; */
+	xmm1 = *(v4sf*)_ps_minus_cephes_DP1;
+	xmm2 = *(v4sf*)_ps_minus_cephes_DP2;
+	xmm3 = *(v4sf*)_ps_minus_cephes_DP3;
+	xmm1 = _mm_mul_ps(y, xmm1);
+	xmm2 = _mm_mul_ps(y, xmm2);
+	xmm3 = _mm_mul_ps(y, xmm3);
+	xx = _mm_add_ps(xx, xmm1);
+	xx = _mm_add_ps(xx, xmm2);
+	xx = _mm_add_ps(xx, xmm3);
+	
+	/* get the sign flag for the cosine */
+	mm4 = _mm_sub_pi32(mm4, *(v2si*)_pi32_2);
+	mm5 = _mm_sub_pi32(mm5, *(v2si*)_pi32_2);
+	mm4 = _mm_andnot_si64(mm4, *(v2si*)_pi32_4);
+	mm5 = _mm_andnot_si64(mm5, *(v2si*)_pi32_4);
+	mm4 = _mm_slli_pi32(mm4, 29);
+	mm5 = _mm_slli_pi32(mm5, 29);
+	v4sf sign_bit_cos;
+	COPY_MM_TO_XMM(mm4, mm5, sign_bit_cos);
+	_mm_empty(); /* good-bye mmx */
+	
+	sign_bit_sin = _mm_xor_ps(sign_bit_sin, swap_sign_bit_sin);
+	
+	
+	/* Evaluate the first polynom  (0 <= x <= Pi/4) */
+	v4sf z = _mm_mul_ps(xx,xx);
+	y = *(v4sf*)_ps_coscof_p0;
+	
+	y = _mm_mul_ps(y, z);
+	y = _mm_add_ps(y, *(v4sf*)_ps_coscof_p1);
+	y = _mm_mul_ps(y, z);
+	y = _mm_add_ps(y, *(v4sf*)_ps_coscof_p2);
+	y = _mm_mul_ps(y, z);
+	y = _mm_mul_ps(y, z);
+	v4sf tmp = _mm_mul_ps(z, *(v4sf*)_ps_0p5);
+	y = _mm_sub_ps(y, tmp);
+	y = _mm_add_ps(y, *(v4sf*)_ps_1);
+	
+	/* Evaluate the second polynom  (Pi/4 <= x <= 0) */
+	
+	v4sf y2 = *(v4sf*)_ps_sincof_p0;
+	y2 = _mm_mul_ps(y2, z);
+	y2 = _mm_add_ps(y2, *(v4sf*)_ps_sincof_p1);
+	y2 = _mm_mul_ps(y2, z);
+	y2 = _mm_add_ps(y2, *(v4sf*)_ps_sincof_p2);
+	y2 = _mm_mul_ps(y2, z);
+	y2 = _mm_mul_ps(y2, xx);
+	y2 = _mm_add_ps(y2, xx);
+	
+	/* select the correct result from the two polynoms */  
+	xmm3 = poly_mask;
+	v4sf ysin2 = _mm_and_ps(xmm3, y2);
+	v4sf ysin1 = _mm_andnot_ps(xmm3, y);
+	y2 = _mm_sub_ps(y2,ysin2);
+	y = _mm_sub_ps(y, ysin1);
+	
+	xmm1 = _mm_add_ps(ysin1,ysin2);
+	xmm2 = _mm_add_ps(y,y2);
+	
+	/* update the sign */
+	_mm_store_ss( s, _mm_xor_ps(xmm1, sign_bit_sin) );
+	_mm_store_ss( c, _mm_xor_ps(xmm2, sign_bit_cos) );
+
 #else
 	#error "Not Implemented"
 #endif
@@ -479,8 +642,102 @@ float _SSE_cos( float x )
 		movss   x,    xmm0
 
 	}
-#elif defined _LINUX || defined __APPLE__
-//	#warning "_SSE_cos NOT implemented!"
+#elif POSIX
+
+	Assert( "Needs testing, verify impl!\n" );
+
+	v4sf xmm1, xmm2 = _mm_setzero_ps(), xmm3, y;
+	v2si mm0, mm1, mm2, mm3;
+	/* take the absolute value */
+	v4sf  xx = _mm_load_ss( &x );
+
+	xx = _mm_and_ps(xx, *(v4sf*)_ps_inv_sign_mask);
+		
+	/* scale by 4/Pi */
+	y = _mm_mul_ps(xx, *(v4sf*)_ps_cephes_FOPI);
+	
+	/* store the integer part of y in mm0:mm1 */
+	xmm2 = _mm_movehl_ps(xmm2, y);
+	mm2 = _mm_cvttps_pi32(y);
+	mm3 = _mm_cvttps_pi32(xmm2);
+	
+	/* j=(j+1) & (~1) (see the cephes sources) */
+	mm2 = _mm_add_pi32(mm2, *(v2si*)_pi32_1);
+	mm3 = _mm_add_pi32(mm3, *(v2si*)_pi32_1);
+	mm2 = _mm_and_si64(mm2, *(v2si*)_pi32_inv1);
+	mm3 = _mm_and_si64(mm3, *(v2si*)_pi32_inv1);
+	
+	y = _mm_cvtpi32x2_ps(mm2, mm3);
+	
+	
+	mm2 = _mm_sub_pi32(mm2, *(v2si*)_pi32_2);
+	mm3 = _mm_sub_pi32(mm3, *(v2si*)_pi32_2);
+	
+	/* get the swap sign flag in mm0:mm1 and the 
+	 polynom selection mask in mm2:mm3 */
+	
+	mm0 = _mm_andnot_si64(mm2, *(v2si*)_pi32_4);
+	mm1 = _mm_andnot_si64(mm3, *(v2si*)_pi32_4);
+	mm0 = _mm_slli_pi32(mm0, 29);
+	mm1 = _mm_slli_pi32(mm1, 29);
+	
+	mm2 = _mm_and_si64(mm2, *(v2si*)_pi32_2);
+	mm3 = _mm_and_si64(mm3, *(v2si*)_pi32_2);
+	
+	mm2 = _mm_cmpeq_pi32(mm2, _mm_setzero_si64());
+	mm3 = _mm_cmpeq_pi32(mm3, _mm_setzero_si64());
+	
+	v4sf sign_bit, poly_mask;
+	COPY_MM_TO_XMM(mm0, mm1, sign_bit);
+	COPY_MM_TO_XMM(mm2, mm3, poly_mask);
+	_mm_empty(); /* good-bye mmx */
+
+	/* The magic pass: "Extended precision modular arithmetic" 
+	 x = ((x - y * DP1) - y * DP2) - y * DP3; */
+	xmm1 = *(v4sf*)_ps_minus_cephes_DP1;
+	xmm2 = *(v4sf*)_ps_minus_cephes_DP2;
+	xmm3 = *(v4sf*)_ps_minus_cephes_DP3;
+	xmm1 = _mm_mul_ps(y, xmm1);
+	xmm2 = _mm_mul_ps(y, xmm2);
+	xmm3 = _mm_mul_ps(y, xmm3);
+	xx = _mm_add_ps(xx, xmm1);
+	xx = _mm_add_ps(xx, xmm2);
+	xx = _mm_add_ps(xx, xmm3);
+	
+	/* Evaluate the first polynom  (0 <= x <= Pi/4) */
+	y = *(v4sf*)_ps_coscof_p0;
+	v4sf z = _mm_mul_ps(xx,xx);
+	
+	y = _mm_mul_ps(y, z);
+	y = _mm_add_ps(y, *(v4sf*)_ps_coscof_p1);
+	y = _mm_mul_ps(y, z);
+	y = _mm_add_ps(y, *(v4sf*)_ps_coscof_p2);
+	y = _mm_mul_ps(y, z);
+	y = _mm_mul_ps(y, z);
+	v4sf tmp = _mm_mul_ps(z, *(v4sf*)_ps_0p5);
+	y = _mm_sub_ps(y, tmp);
+	y = _mm_add_ps(y, *(v4sf*)_ps_1);
+	
+	/* Evaluate the second polynom  (Pi/4 <= x <= 0) */
+	
+	v4sf y2 = *(v4sf*)_ps_sincof_p0;
+	y2 = _mm_mul_ps(y2, z);
+	y2 = _mm_add_ps(y2, *(v4sf*)_ps_sincof_p1);
+	y2 = _mm_mul_ps(y2, z);
+	y2 = _mm_add_ps(y2, *(v4sf*)_ps_sincof_p2);
+	y2 = _mm_mul_ps(y2, z);
+	y2 = _mm_mul_ps(y2, xx);
+	y2 = _mm_add_ps(y2, xx);
+	
+	/* select the correct result from the two polynoms */  
+	xmm3 = poly_mask;
+	y2 = _mm_and_ps(xmm3, y2); //, xmm3);
+	y = _mm_andnot_ps(xmm3, y);
+	y = _mm_add_ps(y,y2);
+	/* update the sign */
+
+	_mm_store_ss( &x, _mm_xor_ps(y, sign_bit) );
+
 #else
 	#error "Not Implemented"
 #endif
@@ -491,6 +748,7 @@ float _SSE_cos( float x )
 //-----------------------------------------------------------------------------
 // SSE2 implementations of optimized routines:
 //-----------------------------------------------------------------------------
+#ifdef PLATFORM_WINDOWS_PC32
 void _SSE2_SinCos(float x, float* s, float* c)  // any x
 {
 #ifdef _WIN32
@@ -569,13 +827,16 @@ void _SSE2_SinCos(float x, float* s, float* c)  // any x
 		movss	[eax], xmm0
 		movss	[edx], xmm6
 	}
-#elif defined _LINUX || defined __APPLE__
-//	#warning "_SSE2_SinCos NOT implemented!"
+#elif POSIX
+	#warning "_SSE2_SinCos NOT implemented!"
+	Assert( 0 );
 #else
 	#error "Not Implemented"
 #endif
 }
+#endif // PLATFORM_WINDOWS_PC32
 
+#ifdef PLATFORM_WINDOWS_PC32
 float _SSE2_cos(float x)  
 {
 #ifdef _WIN32
@@ -624,15 +885,18 @@ float _SSE2_cos(float x)
 		mulss	xmm0, xmm1
 		movss   x,    xmm0
 	}
-#elif defined _LINUX || defined __APPLE__
-//	#warning "_SSE2_cos NOT implemented!"
+#elif POSIX
+	#warning "_SSE2_cos NOT implemented!"
+	Assert( 0 );
 #else
 	#error "Not Implemented"
 #endif
 
 	return x;
 }
+#endif // PLATFORM_WINDOWS_PC32
 
+#if 0
 // SSE Version of VectorTransform
 void VectorTransformSSE(const float *in1, const matrix3x4_t& in2, float *out1)
 {
@@ -681,8 +945,8 @@ void VectorTransformSSE(const float *in1, const matrix3x4_t& in2, float *out1)
 		addss xmm0, [ecx+12]
 		movss [edx+8], xmm0;
 	}
-#elif defined _LINUX || defined __APPLE__
-//	#warning "VectorTransformSSE C implementation only"
+#elif POSIX
+	#warning "VectorTransformSSE C implementation only"
 		out1[0] = DotProduct(in1, in2[0]) + in2[0][3];
 		out1[1] = DotProduct(in1, in2[1]) + in2[1][3];
 		out1[2] = DotProduct(in1, in2[2]) + in2[2][3];
@@ -690,7 +954,9 @@ void VectorTransformSSE(const float *in1, const matrix3x4_t& in2, float *out1)
 	#error "Not Implemented"
 #endif
 }
+#endif
 
+#if 0
 void VectorRotateSSE( const float *in1, const matrix3x4_t& in2, float *out1 )
 {
 	Assert( s_bMathlibInitialized );
@@ -735,8 +1001,8 @@ void VectorRotateSSE( const float *in1, const matrix3x4_t& in2, float *out1 )
 		addss xmm0, xmm2;
 		movss [edx+8], xmm0;
 	}
-#elif defined _LINUX || defined __APPLE__
-//	#warning "VectorRotateSSE C implementation only"
+#elif POSIX
+	#warning "VectorRotateSSE C implementation only"
 		out1[0] = DotProduct( in1, in2[0] );
 		out1[1] = DotProduct( in1, in2[1] );
 		out1[2] = DotProduct( in1, in2[2] );
@@ -744,6 +1010,7 @@ void VectorRotateSSE( const float *in1, const matrix3x4_t& in2, float *out1 )
 	#error "Not Implemented"
 #endif
 }
+#endif
 
 #ifdef _WIN32
 void _declspec(naked) _SSE_VectorMA( const float *start, float scale, const float *direction, float *dest )
@@ -843,3 +1110,4 @@ vec_t DotProduct (const vec_t *a, const vec_t *c)
 }
 */
 
+#endif // COMPILER_MSVC64 

mathlib/sse.h

@@ -15,9 +15,13 @@ void FASTCALL _SSE_VectorNormalizeFast(Vector& vec);
 float _SSE_InvRSquared(const float* v);
 void _SSE_SinCos(float x, float* s, float* c);
 float _SSE_cos( float x);
+#ifdef PLATFORM_WINDOWS_PC32
 void _SSE2_SinCos(float x, float* s, float* c);
 float _SSE2_cos(float x); 
+#endif
+#if 0
 void VectorTransformSSE(const float *in1, const matrix3x4_t& in2, float *out1);
 void VectorRotateSSE( const float *in1, const matrix3x4_t& in2, float *out1 );
+#endif
 
 #endif // _SSE_H

mathlib/sseconst.cpp

@@ -1,4 +1,4 @@
-//===== Copyright © 1996-2005, Valve Corporation, All rights reserved. ======//
+//===== Copyright <EFBFBD> 1996-2005, Valve Corporation, All rights reserved. ======//
 //
 // Purpose: 
 //
@@ -30,6 +30,7 @@ const fltx4 Four_FLT_MAX={FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};
 const fltx4 Four_Negative_FLT_MAX={-FLT_MAX,-FLT_MAX,-FLT_MAX,-FLT_MAX};
 const fltx4 g_SIMD_0123 = { 0., 1., 2., 3. };
 
+extern const fltx4 g_QuatMultRowSign[4];
 const fltx4 g_QuatMultRowSign[4] =
 {
 	{  1.0f,  1.0f, -1.0f, 1.0f },
@@ -38,24 +39,24 @@ const fltx4 g_QuatMultRowSign[4] =
 	{ -1.0f, -1.0f, -1.0f, 1.0f }
 };
 
-const int32 ALIGN16 g_SIMD_clear_signmask[4]= {0x7fffffff,0x7fffffff,0x7fffffff,0x7fffffff};
-const int32 ALIGN16 g_SIMD_signmask[4]= { 0x80000000, 0x80000000, 0x80000000, 0x80000000 };
-const int32 ALIGN16 g_SIMD_lsbmask[4]= { 0xfffffffe, 0xfffffffe, 0xfffffffe, 0xfffffffe };
-const int32 ALIGN16 g_SIMD_clear_wmask[4]= { 0xffffffff, 0xffffffff, 0xffffffff, 0 };
-const int32 ALIGN16 g_SIMD_AllOnesMask[4]= { 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff }; // ~0,~0,~0,~0
-const int32 ALIGN16 g_SIMD_Low16BitsMask[4]= { 0xffff, 0xffff, 0xffff, 0xffff }; // 0xffff x 4
+const int32 ALIGN16 g_SIMD_clear_signmask[4] ALIGN16_POST = {static_cast<int32>(0x7fffffff), static_cast<int32>(0x7fffffff), static_cast<int32>(0x7fffffff), static_cast<int32>(0x7fffffff)};
+const int32 ALIGN16 g_SIMD_signmask[4] ALIGN16_POST = { static_cast<int32>(0x80000000), static_cast<int32>(0x80000000), static_cast<int32>(0x80000000), static_cast<int32>(0x80000000) };
+const int32 ALIGN16 g_SIMD_lsbmask[4] ALIGN16_POST = { static_cast<int32>(0xfffffffe), static_cast<int32>(0xfffffffe), static_cast<int32>(0xfffffffe), static_cast<int32>(0xfffffffe) };
+const int32 ALIGN16 g_SIMD_clear_wmask[4] ALIGN16_POST = { static_cast<int32>(0xffffffff), static_cast<int32>(0xffffffff), static_cast<int32>(0xffffffff), 0 };
+const int32 ALIGN16 g_SIMD_AllOnesMask[4] ALIGN16_POST = { static_cast<int32>(0xffffffff), static_cast<int32>(0xffffffff), static_cast<int32>(0xffffffff), static_cast<int32>(0xffffffff) }; // ~0,~0,~0,~0
+const int32 ALIGN16 g_SIMD_Low16BitsMask[4] ALIGN16_POST = { 0xffff, 0xffff, 0xffff, 0xffff }; // 0xffff x 4
 
-const int32 ALIGN16 g_SIMD_ComponentMask[4][4] =
+const int32 ALIGN16 g_SIMD_ComponentMask[4][4] ALIGN16_POST =
 {
-	{ 0xFFFFFFFF, 0, 0, 0 }, { 0, 0xFFFFFFFF, 0, 0 }, { 0, 0, 0xFFFFFFFF, 0 }, { 0, 0, 0, 0xFFFFFFFF }
+	{ static_cast<int32>(0xFFFFFFFF), 0, 0, 0 }, { 0, static_cast<int32>(0xFFFFFFFF), 0, 0 }, { 0, 0, static_cast<int32>(0xFFFFFFFF), 0 }, { 0, 0, 0, static_cast<int32>(0xFFFFFFFF) }
 };
 
-const int32 ALIGN16 g_SIMD_SkipTailMask[4][4] =
+const int32 ALIGN16 g_SIMD_SkipTailMask[4][4] ALIGN16_POST =
 {
-	{ 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff },
-	{ 0xffffffff, 0x00000000, 0x00000000, 0x00000000 },
-	{ 0xffffffff, 0xffffffff, 0x00000000, 0x00000000 },
-	{ 0xffffffff, 0xffffffff, 0xffffffff, 0x00000000 },
+	{ static_cast<int32>(0xffffffff), static_cast<int32>(0xffffffff), static_cast<int32>(0xffffffff), static_cast<int32>(0xffffffff) },
+	{ static_cast<int32>(0xffffffff), static_cast<int32>(0x00000000), static_cast<int32>(0x00000000), static_cast<int32>(0x00000000) },
+	{ static_cast<int32>(0xffffffff), static_cast<int32>(0xffffffff), static_cast<int32>(0x00000000), static_cast<int32>(0x00000000) },
+	{ static_cast<int32>(0xffffffff), static_cast<int32>(0xffffffff), static_cast<int32>(0xffffffff), static_cast<int32>(0x00000000) },
 };
 
 

mathlib/ssenoise.cpp

@@ -30,6 +30,10 @@ static ALIGN16 int32 idx_mask[4]= {0xffff, 0xffff, 0xffff, 0xffff};
 // returns 0..1
 static inline float GetLatticePointValue( int idx_x, int idx_y, int idx_z )
 {
+	NOTE_UNUSED(perm_d);
+	NOTE_UNUSED(impulse_ycoords);
+	NOTE_UNUSED(impulse_zcoords);
+
 	int ret_idx = perm_a[idx_x & 0xff];
 	ret_idx = perm_b[( idx_y + ret_idx ) & 0xff];
 	ret_idx = perm_c[( idx_z + ret_idx ) & 0xff];

mathlib/vmatrix.cpp

@@ -18,7 +18,7 @@
 // memdbgon must be the last include file in a .cpp file!!!
 #include "tier0/memdbgon.h"
 
-#ifdef _MSC_VER
+#ifdef _WIN32
 #pragma warning (disable : 4700) // local variable 'x' used without having been initialized
 #endif
 
@@ -306,7 +306,7 @@ bool MatrixInverseGeneral(const VMatrix& src, VMatrix& dst)
 	for(iRow=0; iRow < 4; iRow++)
 	{
 		// Find the row with the largest element in this column.
-		fLargest = 0.001f;
+		fLargest = 0.00001f;
 		iLargest = -1;
 		for(iTest=iRow; iTest < 4; iTest++)
 		{
@@ -509,7 +509,7 @@ bool VMatrix::IsRotationMatrix() const
 		FloatMakePositive( v2.Dot(v3) ) < 0.01f;
 }
 
-void VMatrix::SetupMatrixOrgAngles( const Vector &origin, const QAngle &vAngles )
+static void SetupMatrixAnglesInternal( vec_t m[4][4], const QAngle & vAngles )
 {
 	float		sr, sp, sy, cr, cp, cy;
 
@@ -530,6 +530,11 @@ void VMatrix::SetupMatrixOrgAngles( const Vector &origin, const QAngle &vAngles
 	m[0][3] = 0.f;
 	m[1][3] = 0.f;
 	m[2][3] = 0.f;
+}
+
+void VMatrix::SetupMatrixOrgAngles( const Vector &origin, const QAngle &vAngles )
+{
+	SetupMatrixAnglesInternal( m, vAngles );
 	
 	// Add translation
 	m[0][3] = origin.x;
@@ -542,6 +547,21 @@ void VMatrix::SetupMatrixOrgAngles( const Vector &origin, const QAngle &vAngles
 }
 
 
+void	VMatrix::SetupMatrixAngles( const QAngle &vAngles )
+{
+	SetupMatrixAnglesInternal( m, vAngles );
+
+	// Zero everything else
+	m[0][3] = 0.0f;
+	m[1][3] = 0.0f;
+	m[2][3] = 0.0f;
+	m[3][0] = 0.0f;
+	m[3][1] = 0.0f;
+	m[3][2] = 0.0f;
+	m[3][3] = 1.0f;
+}
+
+
 //-----------------------------------------------------------------------------
 // Sets matrix to identity
 //-----------------------------------------------------------------------------
@@ -728,7 +748,7 @@ void Vector4DMultiplyPosition( const VMatrix& src1, Vector const& src2, Vector4D
 {
 	// Make sure it works if src2 == dst
 	Vector tmp;
-	Vector const&v = ( &src2 == &dst.AsVector3D() ) ? tmp : src2;
+	Vector const&v = ( &src2 == &dst.AsVector3D() ) ? static_cast<const Vector&>(tmp) : src2;
 
 	if (&src2 == &dst.AsVector3D())
 	{
@@ -751,7 +771,7 @@ void Vector3DMultiply( const VMatrix &src1, const Vector &src2, Vector &dst )
 {
 	// Make sure it works if src2 == dst
 	Vector tmp;
-	const Vector &v = (&src2 == &dst) ?  tmp : src2;
+	const Vector &v = (&src2 == &dst) ?  static_cast<const Vector&>(tmp) : src2;
 
 	if( &src2 == &dst )
 	{
@@ -772,7 +792,7 @@ void Vector3DMultiplyPositionProjective( const VMatrix& src1, const Vector &src2
 {
 	// Make sure it works if src2 == dst
 	Vector tmp;
-	const Vector &v = (&src2 == &dst) ? tmp: src2;
+	const Vector &v = (&src2 == &dst) ? static_cast<const Vector&>(tmp): src2;
 	if( &src2 == &dst )
 	{
 		VectorCopy( src2, tmp );
@@ -799,7 +819,7 @@ void Vector3DMultiplyProjective( const VMatrix& src1, const Vector &src2, Vector
 {
 	// Make sure it works if src2 == dst
 	Vector tmp;
-	const Vector &v = (&src2 == &dst) ? tmp : src2;
+	const Vector &v = (&src2 == &dst) ? static_cast<const Vector&>(tmp) : src2;
 	if( &src2 == &dst )
 	{
 		VectorCopy( src2, tmp );
@@ -852,7 +872,7 @@ void Vector3DMultiplyTranspose( const VMatrix& src1, const Vector& src2, Vector&
 	bool srcEqualsDst = (&src2 == &dst);
 
 	Vector tmp;
-	const Vector&v = srcEqualsDst ? tmp : src2;
+	const Vector&v = srcEqualsDst ? static_cast<const Vector&>(tmp) : src2;
 
 	if (srcEqualsDst)
 	{
@@ -937,7 +957,7 @@ void MatrixBuildTranslation( VMatrix& dst, const Vector &translation )
 //-----------------------------------------------------------------------------
 void MatrixBuildRotationAboutAxis( VMatrix &dst, const Vector &vAxisOfRot, float angleDegrees )
 {
-	MatrixBuildRotationAboutAxis( vAxisOfRot, angleDegrees, dst.As3x4() );
+	MatrixBuildRotationAboutAxis( vAxisOfRot, angleDegrees, const_cast< matrix3x4_t &> ( dst.As3x4() ) );
 	dst[3][0] = 0;
 	dst[3][1] = 0;
 	dst[3][2] = 0;
@@ -1233,19 +1253,29 @@ void MatrixBuildOrtho( VMatrix& dst, double left, double top, double right, doub
 				0.0f,						0.0f,						0.0f,								1.0f );
 }
 
+void MatrixBuildPerspectiveZRange( VMatrix& dst, double flZNear, double flZFar )
+{
+	dst.m[2][0] = 0.0f;
+	dst.m[2][1] = 0.0f;
+	dst.m[2][2] = flZFar / ( flZNear - flZFar );
+	dst.m[2][3] = flZNear * flZFar / ( flZNear - flZFar );
+}
+
 void MatrixBuildPerspectiveX( VMatrix& dst, double flFovX, double flAspect, double flZNear, double flZFar )
 {
-	float flWidth = 2.0f * flZNear * tanf( flFovX * M_PI / 360.0f );
-	float flHeight = flWidth / flAspect;
-	dst.Init(   2.0f * flZNear / flWidth,						0.0f,							0.0f,										0.0f,
-				0.0f,  2.0f  * flZNear/ flHeight,							0.0f,										0.0f,
-				0.0f,						0.0f,  flZFar / ( flZNear - flZFar ),	 flZNear * flZFar / ( flZNear - flZFar ),
+	float flWidthScale = 1.0f / tanf( flFovX * M_PI / 360.0f );
+	float flHeightScale = flAspect * flWidthScale;
+	dst.Init(   flWidthScale,				0.0f,							0.0f,										0.0f,
+				0.0f,						flHeightScale,					0.0f,										0.0f,
+				0.0f,						0.0f,							0.0f,										0.0f,
 				0.0f,						0.0f,						   -1.0f,										0.0f );
+
+	MatrixBuildPerspectiveZRange ( dst, flZNear, flZFar );
 }
 
 void MatrixBuildPerspectiveOffCenterX( VMatrix& dst, double flFovX, double flAspect, double flZNear, double flZFar, double bottom, double top, double left, double right )
 {
-	float flWidth = 2.0f * flZNear * tanf( flFovX * M_PI / 360.0f );
+	float flWidth = tanf( flFovX * M_PI / 360.0f );
 	float flHeight = flWidth / flAspect;
 
 	// bottom, top, left, right are 0..1 so convert to -<val>/2..<val>/2
@@ -1254,10 +1284,12 @@ void MatrixBuildPerspectiveOffCenterX( VMatrix& dst, double flFovX, double flAsp
 	float flBottom = -(flHeight/2.0f) * (1.0f - bottom) + bottom * (flHeight/2.0f);
 	float flTop    = -(flHeight/2.0f) * (1.0f - top)    + top    * (flHeight/2.0f);
 
-	dst.Init(   (2.0f * flZNear) / (flRight-flLeft),                           0.0f, (flLeft+flRight)/(flRight-flLeft),                            0.0f,
-				0.0f,  2.0f*flZNear/(flTop-flBottom), (flTop+flBottom)/(flTop-flBottom),                            0.0f,
-				0.0f,                           0.0f,           flZFar/(flZNear-flZFar),  flZNear*flZFar/(flZNear-flZFar),
+	dst.Init(   1.0f / (flRight-flLeft),        0.0f,                           (flLeft+flRight)/(flRight-flLeft),  0.0f,
+				0.0f,                           1.0f /(flTop-flBottom),         (flTop+flBottom)/(flTop-flBottom),  0.0f,
+				0.0f,                           0.0f,							0.0f,								0.0f,
 				0.0f,                           0.0f,                           -1.0f,								0.0f );
+
+	MatrixBuildPerspectiveZRange ( dst, flZNear, flZFar );
 }
 #endif // !_STATIC_LINKED || _SHARED_LIB
 

public/appframework/ilaunchermgr.h

@@ -0,0 +1,162 @@
+//========= Copyright Valve Corporation, All rights reserved. ============//
+//
+//	ilaunchermgr.h
+//
+//==================================================================================================
+#ifndef ILAUNCHERMGR_H
+#define ILAUNCHERMGR_H
+
+#ifdef _WIN32
+#pragma once
+#endif
+
+#if defined( USE_SDL )
+
+#include "tier0/threadtools.h"
+#include "appframework/IAppSystem.h"
+
+#if defined( DX_TO_GL_ABSTRACTION )
+#if !defined(DEDICATED)
+#include "togl/linuxwin/glmgrbasics.h"
+#include "togl/linuxwin/glmdisplay.h"
+#endif
+
+class GLMDisplayDB;
+class CShowPixelsParams;
+
+#if defined(DEDICATED)
+typedef void *PseudoGLContextPtr;
+class GLMRendererInfoFields;
+#endif
+#endif
+
+// if you rev this version also update materialsystem/cmaterialsystem.cpp CMaterialSystem::Connect as it defines the string directly
+#define SDLMGR_INTERFACE_VERSION "SDLMgrInterface001"
+
+class CCocoaEvent;
+class CStackCrawlParams;
+
+typedef struct SDL_Cursor SDL_Cursor;
+
+class ILauncherMgr : public IAppSystem
+{
+public:
+	virtual bool Connect( CreateInterfaceFn factory ) = 0;
+	virtual void Disconnect() = 0;
+	
+	virtual void *QueryInterface( const char *pInterfaceName ) = 0;
+	
+	// Init, shutdown
+	virtual InitReturnVal_t Init() = 0;
+	virtual void Shutdown() = 0;
+	
+	// Create the window.
+	virtual bool CreateGameWindow( const char *pTitle, bool bWindowed, int width, int height ) = 0;
+	
+	virtual void IncWindowRefCount() = 0;
+	virtual void DecWindowRefCount() = 0;
+		
+	// Get the next N events. The function returns the number of events that were filled into your array.
+	virtual int GetEvents( CCocoaEvent *pEvents, int nMaxEventsToReturn, bool debugEvents = false ) = 0;
+#ifdef LINUX
+	virtual int PeekAndRemoveKeyboardEvents( bool *pbEsc, bool *pbReturn, bool *pbSpace, bool debugEvents = false ) = 0;
+#endif
+
+	// Set the mouse cursor position.
+	virtual void SetCursorPosition( int x, int y ) = 0;
+	
+	virtual void SetWindowFullScreen( bool bFullScreen, int nWidth, int nHeight ) = 0;
+	virtual bool IsWindowFullScreen() = 0;
+	virtual void MoveWindow( int x, int y ) = 0;
+	virtual void SizeWindow( int width, int tall ) = 0;
+	virtual void PumpWindowsMessageLoop() = 0;
+		
+	virtual void DestroyGameWindow() = 0;
+	virtual void SetApplicationIcon( const char *pchAppIconFile ) = 0;
+	
+	virtual void GetMouseDelta( int &x, int &y, bool bIgnoreNextMouseDelta = false ) = 0;
+
+	virtual void GetNativeDisplayInfo( int nDisplay, uint &nWidth, uint &nHeight, uint &nRefreshHz ) = 0; // Retrieve the size of the monitor (desktop)
+	virtual void RenderedSize( uint &width, uint &height, bool set ) = 0;	// either set or retrieve rendered size value (from dxabstract)
+	virtual void DisplayedSize( uint &width, uint &height ) = 0;			// query backbuffer size (window size whether FS or windowed)
+	
+#if defined( DX_TO_GL_ABSTRACTION )
+	virtual PseudoGLContextPtr	GetMainContext() = 0;
+	// Get the NSGLContext for a window's main view - note this is the carbon windowref as an argument
+	virtual PseudoGLContextPtr GetGLContextForWindow( void* windowref ) = 0;
+	virtual PseudoGLContextPtr CreateExtraContext() = 0;
+	virtual void DeleteContext( PseudoGLContextPtr hContext ) = 0;
+	virtual bool MakeContextCurrent( PseudoGLContextPtr hContext ) = 0;
+	virtual GLMDisplayDB *GetDisplayDB( void ) = 0;
+	virtual void GetDesiredPixelFormatAttribsAndRendererInfo( uint **ptrOut, uint *countOut, GLMRendererInfoFields *rendInfoOut ) = 0;
+	virtual void ShowPixels( CShowPixelsParams *params ) = 0;
+#endif
+
+	virtual void GetStackCrawl( CStackCrawlParams *params ) = 0;	
+	
+	virtual void WaitUntilUserInput( int msSleepTime ) = 0;
+	
+	virtual void *GetWindowRef() = 0;
+
+	virtual void SetMouseVisible( bool bState ) = 0;
+	virtual void SetMouseCursor( SDL_Cursor *hCursor ) = 0;
+	virtual void SetForbidMouseGrab( bool bForbidMouseGrab ) = 0;
+	virtual void OnFrameRendered() = 0;
+
+	virtual void SetGammaRamp( const uint16 *pRed, const uint16 *pGreen, const uint16 *pBlue ) = 0;
+
+	virtual double GetPrevGLSwapWindowTime() = 0;
+};
+
+extern ILauncherMgr *g_pLauncherMgr;
+
+enum CocoaEventType_t
+{
+	CocoaEvent_KeyDown,
+	CocoaEvent_KeyUp,
+	CocoaEvent_MouseButtonDown,
+	CocoaEvent_MouseMove,
+	CocoaEvent_MouseButtonUp,
+	CocoaEvent_AppActivate,
+	CocoaEvent_MouseScroll,
+	CocoaEvent_AppQuit,
+	CocoaEvent_Deleted, // Event was one of the above, but has been handled and should be ignored now.
+};
+
+// enum values need to match bit-shifting logic in CInputSystem::UpdateMouseButtonState and 
+// the codes from NSEvent pressedMouseButtons, turns out the two are in agreement right now
+enum CocoaMouseButton_t
+{
+	COCOABUTTON_LEFT = 1 << 0,
+	COCOABUTTON_RIGHT = 1 << 1,
+	COCOABUTTON_MIDDLE = 1 << 2,
+	COCOABUTTON_4 = 1 << 3,	
+	COCOABUTTON_5 = 1 << 4,	
+};
+
+enum ECocoaKeyModifier
+{
+	eCapsLockKey,
+	eShiftKey,
+	eControlKey,
+	eAltKey,		// aka option
+	eCommandKey
+};
+
+class CCocoaEvent
+{
+public:
+	CocoaEventType_t m_EventType;
+	int m_VirtualKeyCode;
+	wchar_t m_UnicodeKey;
+	wchar_t m_UnicodeKeyUnmodified;
+	uint m_ModifierKeyMask;		// 
+	int m_MousePos[2];
+	int m_MouseButtonFlags;	// Current state of the mouse buttons. See COCOABUTTON_xxxx.
+	uint m_nMouseClickCount;
+	int m_MouseButton; // which of the CocoaMouseButton_t buttons this is for from above
+};
+
+#endif // defined( USE_SDL )
+
+#endif // ILAUNCHERMGR_H

public/basehandle.h

@@ -1,4 +1,4 @@
-//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
+//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
 //
 // Purpose: 
 //
@@ -30,7 +30,7 @@ public:
 
 	CBaseHandle();
 	CBaseHandle( const CBaseHandle &other );
-	CBaseHandle( unsigned long value );
+	CBaseHandle( uint32_t value );
 	CBaseHandle( int iEntry, int iSerialNumber );
 
 	void Init( int iEntry, int iSerialNumber );
@@ -63,7 +63,7 @@ public:
 protected:
 	// The low NUM_SERIAL_BITS hold the index. If this value is less than MAX_EDICTS, then the entity is networkable.
 	// The high NUM_SERIAL_NUM_BITS bits are the serial number.
-	unsigned long	m_Index;
+	uint32_t	m_Index;
 };
 
 
@@ -80,7 +80,7 @@ inline CBaseHandle::CBaseHandle( const CBaseHandle &other )
 	m_Index = other.m_Index;
 }
 
-inline CBaseHandle::CBaseHandle( unsigned long value )
+inline CBaseHandle::CBaseHandle( uint32_t value )
 {
 	m_Index = value;
 }
@@ -92,10 +92,10 @@ inline CBaseHandle::CBaseHandle( int iEntry, int iSerialNumber )
 
 inline void CBaseHandle::Init( int iEntry, int iSerialNumber )
 {
-	Assert( iEntry >= 0 && iEntry < NUM_ENT_ENTRIES );
+	Assert( iEntry >= 0 && (iEntry & ENT_ENTRY_MASK) == iEntry);
 	Assert( iSerialNumber >= 0 && iSerialNumber < (1 << NUM_SERIAL_NUM_BITS) );
 
-	m_Index = iEntry | (iSerialNumber << NUM_ENT_ENTRY_BITS);
+	m_Index = iEntry | (iSerialNumber << NUM_SERIAL_NUM_SHIFT_BITS);
 }
 
 inline void CBaseHandle::Term()
@@ -110,12 +110,26 @@ inline bool CBaseHandle::IsValid() const
 
 inline int CBaseHandle::GetEntryIndex() const
 {
+	// There is a hack here: due to a bug in the original implementation of the 
+	// entity handle system, an attempt to look up an invalid entity index in 
+	// certain cirumstances might fall through to the the mask operation below.
+	// This would mask an invalid index to be in fact a lookup of entity number
+	// NUM_ENT_ENTRIES, so invalid ent indexes end up actually looking up the
+	// last slot in the entities array. Since this slot is always empty, the 
+	// lookup returns NULL and the expected behavior occurs through this unexpected
+	// route.
+	// A lot of code actually depends on this behavior, and the bug was only exposed
+	// after a change to NUM_SERIAL_NUM_BITS increased the number of allowable
+	// static props in the world. So the if-stanza below detects this case and 
+	// retains the prior (bug-submarining) behavior.
+	if ( !IsValid() )
+		return NUM_ENT_ENTRIES-1;
 	return m_Index & ENT_ENTRY_MASK;
 }
 
 inline int CBaseHandle::GetSerialNumber() const
 {
-	return m_Index >> NUM_ENT_ENTRY_BITS;
+	return m_Index >> NUM_SERIAL_NUM_SHIFT_BITS;
 }
 
 inline int CBaseHandle::ToInt() const
@@ -150,7 +164,7 @@ inline bool CBaseHandle::operator <( const CBaseHandle &other ) const
 
 inline bool CBaseHandle::operator <( const IHandleEntity *pEntity ) const
 {
-	unsigned long otherIndex = (pEntity) ? pEntity->GetRefEHandle().m_Index : INVALID_EHANDLE_INDEX;
+	uint32_t otherIndex = (pEntity) ? pEntity->GetRefEHandle().m_Index : INVALID_EHANDLE_INDEX;
 	return m_Index < otherIndex;
 }
 

public/bitvec.h

@@ -1403,7 +1403,7 @@ inline void CBitVecAccessor::operator=(int val)
 	if(val)
 		m_pDWords[m_iBit >> 5] |= (1 << (m_iBit & 31));
 	else
-		m_pDWords[m_iBit >> 5] &= ~(unsigned long)(1 << (m_iBit & 31));
+		m_pDWords[m_iBit >> 5] &= ~(uint32_t)(1 << (m_iBit & 31));
 }
 
 inline CBitVecAccessor::operator uint32()

public/bspfile.h

@@ -660,7 +660,7 @@ public:
 	CDispCornerNeighbors	m_CornerNeighbors[4];	// Indexed by CORNER_ defines.
 
 	enum unnamed { ALLOWEDVERTS_SIZE = PAD_NUMBER( MAX_DISPVERTS, 32 ) / 32 };
-	unsigned long	m_AllowedVerts[ALLOWEDVERTS_SIZE];	// This is built based on the layout and sizes of our neighbors
+	uint32_t	m_AllowedVerts[ALLOWEDVERTS_SIZE];	// This is built based on the layout and sizes of our neighbors
 														// and tells us which vertices are allowed to be active.
 };
 

public/const.h

@@ -1,4 +1,4 @@
-//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
+//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
 //
 // Purpose: 
 //
@@ -63,13 +63,13 @@
 #define SIGNED_GUID_LEN 32 // Hashed CD Key (32 hex alphabetic chars + 0 terminator )
 
 // Used for networking ehandles.
-#define NUM_ENT_ENTRY_BITS		(MAX_EDICT_BITS + 1)
+#define NUM_ENT_ENTRY_BITS		(MAX_EDICT_BITS + 2)
 #define NUM_ENT_ENTRIES			(1 << NUM_ENT_ENTRY_BITS)
 #define ENT_ENTRY_MASK			(NUM_ENT_ENTRIES - 1)
 #define INVALID_EHANDLE_INDEX	0xFFFFFFFF
 
 #define NUM_SERIAL_NUM_BITS		(32 - NUM_ENT_ENTRY_BITS)
-
+#define NUM_SERIAL_NUM_SHIFT_BITS       16
 
 // Networked ehandles use less bits to encode the serial number.
 #define NUM_NETWORKED_EHANDLE_SERIAL_NUMBER_BITS	10

public/datacache/idatacache.h

@@ -513,13 +513,15 @@ public:
 		case DC_AGE_DISCARD:
 		case DC_FLUSH_DISCARD:
 		case DC_REMOVED:
+			{
 				STORAGE_TYPE *p = (STORAGE_TYPE *)notification.clientId;
 				p->DestroyResource();
-			return true;
 			}
-
+			return true;
+		default:
 			return CDefaultDataCacheClient::HandleCacheNotification( notification );
 		}
+	}
 
 
 protected:

public/eiface.h

@@ -1,4 +1,4 @@
-//===== Copyright © 1996-2005, Valve Corporation, All rights reserved. ======//
+//===== Copyright © 1996-2005, Valve Corporation, All rights reserved. ======//
 //
 // Purpose: 
 //
@@ -60,6 +60,7 @@ class CSteamID;
 class IReplayFactory;
 class IReplaySystem;
 class IServer;
+class WorkshopMapDesc_t;
 
 typedef struct player_info_s player_info_t;
 
@@ -73,6 +74,8 @@ typedef struct player_info_s player_info_t;
 #define DLLEXPORT /* */
 #endif
 
+#define INTERFACEVERSION_VENGINESERVER_VERSION_21	"VEngineServer021"
+#define INTERFACEVERSION_VENGINESERVER_VERSION_22	"VEngineServer022"
 #define INTERFACEVERSION_VENGINESERVER				"VEngineServer023"
 #define INTERFACEVERSION_VENGINESERVER_INT			23
 
@@ -340,9 +343,6 @@ public:
 	// even though we may not have waited enough time
 	virtual void			AllowImmediateEdictReuse( ) = 0;
 
-	// Returns true if the engine is an internal build. i.e. is using the internal bugreporter.
-	virtual bool		IsInternalBuild( void ) = 0;
-
 	virtual IChangeInfoAccessor *GetChangeAccessor( const edict_t *pEdict ) = 0;	
 
 	// Name of most recently load .sav file
@@ -410,8 +410,6 @@ public:
 	// Server version from the steam.inf, this will be compared to the GC version
 	virtual int GetServerVersion() const = 0;
 
-	// Only in VEngineServer022 and later
-	
 	// Get sv.GetTime()
 	virtual float GetServerTime() const = 0;
 
@@ -447,14 +445,15 @@ public:
 	virtual IReplaySystem *GetReplay() = 0;
 };
 
-#define INTERFACEVERSION_SERVERGAMEDLL_VERSION_4	"ServerGameDLL004"
-#define INTERFACEVERSION_SERVERGAMEDLL_VERSION_5	"ServerGameDLL005"
-#define INTERFACEVERSION_SERVERGAMEDLL_VERSION_6	"ServerGameDLL006"
-#define INTERFACEVERSION_SERVERGAMEDLL_VERSION_7	"ServerGameDLL007"
+// These only differ in new items added to the end
+typedef IVEngineServer IVEngineServer021;
+typedef IVEngineServer IVEngineServer022;
+
 #define INTERFACEVERSION_SERVERGAMEDLL_VERSION_8	"ServerGameDLL008"
 #define INTERFACEVERSION_SERVERGAMEDLL_VERSION_9	"ServerGameDLL009"
-#define INTERFACEVERSION_SERVERGAMEDLL				"ServerGameDLL010"
-#define INTERFACEVERSION_SERVERGAMEDLL_INT			10
+#define INTERFACEVERSION_SERVERGAMEDLL_VERSION_10	"ServerGameDLL010"
+#define INTERFACEVERSION_SERVERGAMEDLL				"ServerGameDLL012"
+#define INTERFACEVERSION_SERVERGAMEDLL_INT			12
 
 class IServerGCLobby;
 
@@ -630,6 +629,8 @@ public:
 
 	// Called to see if the game server is okay with a manual changelevel or map command
 	virtual bool			IsManualMapChangeOkay( const char **pszReason ) = 0;
+
+	virtual bool GetWorkshopMap( unsigned int unk, WorkshopMapDesc_t *pMapDesc ) = 0;
 };
 
 typedef IServerGameDLL IServerGameDLL008;

public/engine/ivdebugoverlay.h

@@ -39,6 +39,7 @@ public:
 	virtual void AddTextOverlay(const Vector& origin, float duration, const char *format, ...) = 0;
 	virtual void AddTextOverlay(const Vector& origin, int line_offset, float duration, const char *format, ...) = 0;
 	virtual void AddScreenTextOverlay(float flXPos, float flYPos,float flDuration, int r, int g, int b, int a, const char *text) = 0;
+	virtual void AddScreenTextOverlay(float flXPos, float flYPos, int line_offset, float flDuration, int r, int g, int b, int a, const char *text) = 0;
 	virtual void AddSweptBoxOverlay(const Vector& start, const Vector& end, const Vector& mins, const Vector& max, const QAngle & angles, int r, int g, int b, int a, float flDuration) = 0;
 	virtual void AddGridOverlay(const Vector& origin) = 0;
 	virtual int ScreenPosition(const Vector& point, Vector& screen) = 0;

public/filesystem.h

@@ -420,7 +420,7 @@ public:
 	virtual bool			IsFileWritable( char const *pFileName, const char *pPathID = 0 ) = 0;
 	virtual bool			SetFileWritable( char const *pFileName, bool writable, const char *pPathID = 0 ) = 0;
 
-	virtual long			GetFileTime( const char *pFileName, const char *pPathID = 0 ) = 0;
+	virtual int32_t			GetFileTime( const char *pFileName, const char *pPathID = 0 ) = 0;
 
 	//--------------------------------------------------------
 	// Reads/writes files to utlbuffers. Use this for optimal read performance when doing open/read/close
@@ -505,7 +505,7 @@ public:
 	// File I/O and info
 	virtual bool			IsDirectory( const char *pFileName, const char *pathID = 0 ) = 0;
 
-	virtual void			FileTimeToString( char* pStrip, int maxCharsIncludingTerminator, long fileTime ) = 0;
+	virtual void			FileTimeToString( char* pStrip, int maxCharsIncludingTerminator, int32_t fileTime ) = 0;
 
 	//--------------------------------------------------------
 	// Open file operations
@@ -716,7 +716,7 @@ public:
 	virtual bool		FullPathToRelativePathEx( const char *pFullpath, const char *pPathId, char *pRelative, int maxlen ) = 0;
 
 	virtual int			GetPathIndex( const FileNameHandle_t &handle ) = 0;
-	virtual long		GetPathTime( const char *pPath, const char *pPathID ) = 0;
+	virtual int32_t		GetPathTime( const char *pPath, const char *pPathID ) = 0;
 
 	virtual DVDMode_t	GetDVDMode() = 0;
 

public/filesystem/IQueuedLoader.h

@@ -14,6 +14,8 @@
 #include "tier0/platform.h"
 #include "appframework/IAppSystem.h"
 
+class CFunctor;
+
 enum LoaderError_t
 {
 	LOADERERROR_NONE     = 0,
@@ -30,6 +32,8 @@ enum LoaderPriority_t
 
 typedef void ( *QueuedLoaderCallback_t )( void *pContext, void *pContext2, const void *pData, int nSize, LoaderError_t loaderError );
 
+typedef void ( *DynamicResourceCallback_t )( const char *pFilename, void *pContext, void *pContext2 );
+
 struct LoaderJob_t
 {
 	LoaderJob_t()
@@ -106,7 +110,7 @@ public:
 #define LOADER_DETAIL_LATECOMPLETIONS	(1<<2)
 #define LOADER_DETAIL_PURGES			(1<<3)
 
-#define QUEUEDLOADER_INTERFACE_VERSION		"QueuedLoaderVersion001"
+#define QUEUEDLOADER_INTERFACE_VERSION		"QueuedLoaderVersion004"
 abstract_class IQueuedLoader : public IAppSystem
 {
 public:
@@ -122,6 +126,11 @@ public:
 	// injects a resource into the map's reslist, rejected if not understood
 	virtual void				AddMapResource( const char *pFilename ) = 0;
 
+	// dynamically load a map resource
+	virtual void				DynamicLoadMapResource( const char *pFilename, DynamicResourceCallback_t pCallback, void *pContext, void *pContext2 ) = 0;
+	virtual void				QueueDynamicLoadFunctor( CFunctor* pFunctor ) = 0;
+	virtual bool				CompleteDynamicLoad() = 0;
+
 	// callback is asynchronous
 	virtual bool				ClaimAnonymousJob( const char *pFilename, QueuedLoaderCallback_t pCallback, void *pContext, void *pContext2 = NULL ) = 0;
 	// provides data if loaded, caller owns data
@@ -134,6 +143,8 @@ public:
 	// callers can expect that jobs are not immediately started when batching
 	virtual bool				IsBatching() const = 0;
 
+	virtual bool				IsDynamic() const = 0;
+
 	// callers can conditionalize operational spew
 	virtual int					GetSpewDetail() const = 0;
 

public/filesystem_init.cpp

@@ -1,4 +1,4 @@
-//====== Copyright © 1996-2004, Valve Corporation, All rights reserved. =======
+//========= Copyright Valve Corporation, All rights reserved. ============//
 //
 // Purpose: 
 //
@@ -6,21 +6,24 @@
 
 #undef PROTECTED_THINGS_ENABLE
 #undef PROTECT_FILEIO_FUNCTIONS
+#ifndef POSIX
+#undef fopen
+#endif
 
 #if defined( _WIN32 ) && !defined( _X360 )
 #include <windows.h>
 #include <direct.h>
 #include <io.h>
 #include <process.h>
-#elif defined( _LINUX ) || defined( __APPLE__ )
+#elif defined( POSIX )
 #include <unistd.h>
-#define _putenv putenv
 #define _chdir chdir
 #define _access access
 #endif
 #include <stdio.h>
 #include <sys/stat.h>
 #include "tier1/strtools.h"
+#include "tier1/utlbuffer.h"
 #include "filesystem_init.h"
 #include "tier0/icommandline.h"
 #include "KeyValues.h"
@@ -83,7 +86,7 @@ public:
 		if ( pValue )
 		{
 			m_bExisted = true;
-			m_OriginalValue.SetSize( strlen( pValue ) + 1 );
+			m_OriginalValue.SetSize( Q_strlen( pValue ) + 1 );
 			memcpy( m_OriginalValue.Base(), pValue, m_OriginalValue.Count() );
 		}
 		else
@@ -144,16 +147,24 @@ public:
 		Q_vsnprintf( valueString, sizeof( valueString ), pValue, marker );
 		va_end( marker );
 
+#ifdef WIN32
 		char str[4096];
 		Q_snprintf( str, sizeof( str ), "%s=%s", m_pVarName, valueString );
 		_putenv( str );
+#else
+		setenv( m_pVarName, valueString, 1 );
+#endif
 	}
 
 	void ClearValue()
 	{
+#ifdef WIN32
 		char str[512];
 		Q_snprintf( str, sizeof( str ), "%s=", m_pVarName );
 		_putenv( str );
+#else
+		setenv( m_pVarName, "", 1 );
+#endif
 	}
 
 private:
@@ -213,6 +224,7 @@ CFSSearchPathsInit::CFSSearchPathsInit()
 	m_pDirectoryName = NULL;
 	m_pLanguage = NULL;
 	m_ModPath[0] = 0;
+	m_bMountHDContent = m_bLowViolence = false;
 }
 
 
@@ -248,47 +260,6 @@ const char *FileSystem_GetLastErrorString()
 }
 
 
-void AddLanguageGameDir( IFileSystem *pFileSystem, const char *pLocation, const char *pLanguage )
-{
-	if ( IsX360() )
-	{
-		// 360 does not use this path for localization
-		return;
-	}
-
-#if !defined( SWDS )
-	char temp[MAX_PATH];
-	Q_snprintf( temp, sizeof(temp), "%s_%s", pLocation, pLanguage );
-	pFileSystem->AddSearchPath( temp, "GAME", PATH_ADD_TO_TAIL );
-
-	if ( !pFileSystem->IsSteam() )
-	{
-		// also look in "..\localization\<folder>" if not running Steam
-		char baseDir[MAX_PATH];
-		char *tempPtr = NULL, *gameDir = NULL;
-
-		Q_strncpy( baseDir, pLocation, sizeof(baseDir) );
-		tempPtr = Q_strstr( baseDir, "\\game\\" );
-		
-		if ( tempPtr )
-		{
-			gameDir = tempPtr + Q_strlen( "\\game\\" );
-			*tempPtr = 0;
-			Q_snprintf( temp, sizeof(temp), "%s%clocalization%c%s_%s", baseDir, CORRECT_PATH_SEPARATOR, CORRECT_PATH_SEPARATOR, gameDir, pLanguage );
-			pFileSystem->AddSearchPath( temp, "GAME", PATH_ADD_TO_TAIL );
-		}
-	}
-#endif
-}
-
-
-void AddGameBinDir( IFileSystem *pFileSystem, const char *pLocation )
-{
-	char temp[MAX_PATH];
-	Q_snprintf( temp, sizeof(temp), "%s%cbin", pLocation, CORRECT_PATH_SEPARATOR );
-	pFileSystem->AddSearchPath( temp, "GAMEBIN", PATH_ADD_TO_TAIL );
-}
-
 KeyValues* ReadKeyValuesFile( const char *pFilename )
 {
 	// Read in the gameinfo.txt file and null-terminate it.
@@ -376,7 +347,8 @@ bool FileSystem_GetExecutableDir( char *exedir, int exeDirLen )
 	Q_StrRight( exedir, 4, ext, sizeof( ext ) );
 	if ( ext[0] != CORRECT_PATH_SEPARATOR || Q_stricmp( ext+1, "bin" ) != 0 )
 	{
-		Q_strncat( exedir, "\\bin", exeDirLen, COPY_ALL_CHARACTERS );
+		Q_strncat( exedir, CORRECT_PATH_SEPARATOR_S, exeDirLen, COPY_ALL_CHARACTERS );
+		Q_strncat( exedir, "bin", exeDirLen, COPY_ALL_CHARACTERS );
 		Q_FixSlashes( exedir );
 	}
 	
@@ -489,124 +461,73 @@ FSReturnCode_t LoadGameInfoFile(
 	return FS_OK;
 }
 
-// checks the registry for the low violence setting
-// Check "HKEY_CURRENT_USER\Software\Valve\Source\Settings" and "User Token 2" or "User Token 3"
-bool IsLowViolenceBuild( void )
-{
-#if defined(_WIN32)
-	HKEY hKey;
-	char szValue[64];
-	unsigned long len = sizeof(szValue) - 1;
-	bool retVal = false;
-	
-	if ( IsPC() && RegOpenKeyEx( HKEY_CURRENT_USER, "Software\\Valve\\Source\\Settings", NULL, KEY_READ, &hKey) == ERROR_SUCCESS )
-	{
-		// User Token 2
-		if ( RegQueryValueEx( hKey, "User Token 2", NULL, NULL, (unsigned char*)szValue, &len ) == ERROR_SUCCESS )
-		{
-			if ( Q_strlen( szValue ) > 0 )
-			{
-				retVal = true;
-			}
-		}
-
-		if ( !retVal )
-		{
-			// reset "len" for the next check
-			len = sizeof(szValue) - 1;
-
-			// User Token 3
-			if ( RegQueryValueEx( hKey, "User Token 3", NULL, NULL, (unsigned char*)szValue, &len ) == ERROR_SUCCESS )
-			{
-				if ( Q_strlen( szValue ) > 0 )
-				{
-					retVal = true;
-				}
-			}
-		}
-
-		RegCloseKey(hKey);
-	}
-
-	return retVal;
-#elif _LINUX
-	return false;
-#else
-	#error "Fix me"
-#endif
-}
 
 static void FileSystem_AddLoadedSearchPath( 
 	CFSSearchPathsInit &initInfo, 
 	const char *pPathID, 
-	bool *bFirstGamePath, 
-	const char *pBaseDir, 
-	const char *pLocation,
+	const char *fullLocationPath, 
 	bool bLowViolence )
 {
-	char fullLocationPath[MAX_PATH];
-	Q_MakeAbsolutePath( fullLocationPath, sizeof( fullLocationPath ), pLocation, pBaseDir );
 
-	// Now resolve any ./'s.
-	V_FixSlashes( fullLocationPath );
-	if ( !V_RemoveDotSlashes( fullLocationPath ) )
-		Error( "FileSystem_AddLoadedSearchPath - Can't resolve pathname for '%s'", fullLocationPath );
+	// Check for mounting LV game content in LV builds only
+	if ( V_stricmp( pPathID, "game_lv" ) == 0 )
+	{
 
-	// Add language, mod, and gamebin search paths automatically.
-	if ( Q_stricmp( pPathID, "game" ) == 0 )
-	{
-		// add the low violence path
-		if ( bLowViolence )
-		{
-			char szPath[MAX_PATH];
-			Q_snprintf( szPath, sizeof(szPath), "%s_lv", fullLocationPath );
-			initInfo.m_pFileSystem->AddSearchPath( szPath, pPathID, PATH_ADD_TO_TAIL );
+		// Not in LV build, don't mount
+		if ( !initInfo.m_bLowViolence )
+			return;
+
+		// Mount, as a game path
+		pPathID = "game";
 	}
 
-		// add the language path
-		if ( initInfo.m_pLanguage )
+	// Check for mounting HD game content if enabled
+	if ( V_stricmp( pPathID, "game_hd" ) == 0 )
 	{
-			AddLanguageGameDir( initInfo.m_pFileSystem, fullLocationPath, initInfo.m_pLanguage );
+
+		// Not in LV build, don't mount
+		if ( !initInfo.m_bMountHDContent )
+			return;
+
+		// Mount, as a game path
+		pPathID = "game";
 	}
 
+
+	// Special processing for ordinary game folders
+	if ( V_stristr( fullLocationPath, ".vpk" ) == NULL && Q_stricmp( pPathID, "game" ) == 0 )
+	{
 		if ( CommandLine()->FindParm( "-tempcontent" ) != 0 )
 		{
 			char szPath[MAX_PATH];
 			Q_snprintf( szPath, sizeof(szPath), "%s_tempcontent", fullLocationPath );
 			initInfo.m_pFileSystem->AddSearchPath( szPath, pPathID, PATH_ADD_TO_TAIL );
 		}
-
-		// mark the first "game" dir as the "MOD" dir
-		if ( *bFirstGamePath )
-		{
-			*bFirstGamePath = false;
-			initInfo.m_pFileSystem->AddSearchPath( fullLocationPath, "MOD", PATH_ADD_TO_TAIL );
-			Q_strncpy( initInfo.m_ModPath, fullLocationPath, sizeof( initInfo.m_ModPath ) );
 	}
 
-		// add the game bin
-		AddGameBinDir( initInfo.m_pFileSystem, fullLocationPath );
+	
+	if ( initInfo.m_pLanguage &&
+	     Q_stricmp( initInfo.m_pLanguage, "english" ) &&
+	     V_strstr( fullLocationPath, "_english" ) != NULL )
+	{
+		char szPath[MAX_PATH];
+		char szLangString[MAX_PATH];		
+		
+		// Need to add a language version of this path first
+
+		Q_snprintf( szLangString, sizeof(szLangString), "_%s", initInfo.m_pLanguage);
+		V_StrSubst( fullLocationPath, "_english", szLangString, szPath, sizeof( szPath ), true );
+		initInfo.m_pFileSystem->AddSearchPath( szPath, pPathID, PATH_ADD_TO_TAIL );		
 	}
 
 	initInfo.m_pFileSystem->AddSearchPath( fullLocationPath, pPathID, PATH_ADD_TO_TAIL );
 }
 
-
-bool FileSystem_IsHldsUpdateToolDedicatedServer()
+static int SortStricmp( char * const * sz1, char * const * sz2 )
 {
-	// To determine this, we see if the directory our executable was launched from is "orangebox".
-	// We only are under "orangebox" if we're run from hldsupdatetool.
-	char baseDir[MAX_PATH];
-	if ( !FileSystem_GetBaseDir( baseDir, sizeof( baseDir ) ) )
-		return false;
-
-	V_FixSlashes( baseDir );
-	V_StripTrailingSlash( baseDir );
-	const char *pLastDir = V_UnqualifiedFileName( baseDir );
-	return ( pLastDir && V_stricmp( pLastDir, "orangebox" ) == 0 );
+	return V_stricmp( *sz1, *sz2 );
 }
 
-
 FSReturnCode_t FileSystem_LoadSearchPaths( CFSSearchPathsInit &initInfo )
 {
 	if ( !initInfo.m_pFileSystem || !initInfo.m_pDirectoryName )
@@ -622,23 +543,41 @@ FSReturnCode_t FileSystem_LoadSearchPaths( CFSSearchPathsInit &initInfo )
 	if ( !FileSystem_GetBaseDir( baseDir, sizeof( baseDir ) ) )
 		return SetupFileSystemError( false, FS_INVALID_PARAMETERS, "FileSystem_GetBaseDir failed." );
 
-	initInfo.m_ModPath[0] = 0;
+	// The MOD directory is always the one that contains gameinfo.txt
+	Q_strncpy( initInfo.m_ModPath, initInfo.m_pDirectoryName, sizeof( initInfo.m_ModPath ) );
 
 	#define GAMEINFOPATH_TOKEN		"|gameinfo_path|"
 	#define BASESOURCEPATHS_TOKEN	"|all_source_engine_paths|"
 
-	bool bLowViolence = IsLowViolenceBuild();
-	bool bFirstGamePath = true;
+	const char *pszExtraSearchPath = CommandLine()->ParmValue( "-insert_search_path" );
+	if ( pszExtraSearchPath )
+	{
+		CUtlStringList vecPaths;
+		V_SplitString( pszExtraSearchPath, ",", vecPaths );
+		FOR_EACH_VEC( vecPaths, idxExtraPath )
+		{
+			char szAbsSearchPath[MAX_PATH];
+			Q_StripPrecedingAndTrailingWhitespace( vecPaths[ idxExtraPath ] );
+			V_MakeAbsolutePath( szAbsSearchPath, sizeof( szAbsSearchPath ), vecPaths[ idxExtraPath ], baseDir );
+			V_FixSlashes( szAbsSearchPath );
+			if ( !V_RemoveDotSlashes( szAbsSearchPath ) )
+				Error( "Bad -insert_search_path - Can't resolve pathname for '%s'", szAbsSearchPath );
+			V_StripTrailingSlash( szAbsSearchPath );
+			FileSystem_AddLoadedSearchPath( initInfo, "GAME", szAbsSearchPath, false );
+			FileSystem_AddLoadedSearchPath( initInfo, "MOD", szAbsSearchPath, false );
+		}
+	}
 
+	bool bLowViolence = initInfo.m_bLowViolence;
 	for ( KeyValues *pCur=pSearchPaths->GetFirstValue(); pCur; pCur=pCur->GetNextValue() )
 	{
-		const char *pPathID = pCur->GetName();
 		const char *pLocation = pCur->GetString();
+		const char *pszBaseDir = baseDir;
 
 		if ( Q_stristr( pLocation, GAMEINFOPATH_TOKEN ) == pLocation )
 		{
 			pLocation += strlen( GAMEINFOPATH_TOKEN );
-			FileSystem_AddLoadedSearchPath( initInfo, pPathID, &bFirstGamePath, initInfo.m_pDirectoryName, pLocation, bLowViolence );
+			pszBaseDir = initInfo.m_pDirectoryName;
 		}
 		else if ( Q_stristr( pLocation, BASESOURCEPATHS_TOKEN ) == pLocation )
 		{
@@ -650,26 +589,117 @@ FSReturnCode_t FileSystem_LoadSearchPaths( CFSSearchPathsInit &initInfo )
 			// We need a special identifier in the gameinfo.txt here because the base hl2 folder exists in different places.
 			// In the case of a game or a Steam-launched dedicated server, all the necessary prior engine content is mapped in with the Steam depots,
 			// so we can just use the path as-is.
-
-			// In the case of an hldsupdatetool dedicated server, the base hl2 folder is "..\..\hl2" (since we're up in the 'orangebox' folder).
-												   
 			pLocation += strlen( BASESOURCEPATHS_TOKEN );
-
-			// Add the Orange-box path (which also will include whatever the depots mapped in as well if we're 
-			// running a Steam-launched app).
-			FileSystem_AddLoadedSearchPath( initInfo, pPathID, &bFirstGamePath, baseDir, pLocation, bLowViolence );
-
-			if ( FileSystem_IsHldsUpdateToolDedicatedServer() )
-			{			
-				// If we're using the hldsupdatetool dedicated server, then go up a directory to get the ep1-era files too.
-				char ep1EraPath[MAX_PATH];
-				V_snprintf( ep1EraPath, sizeof( ep1EraPath ), "..%c%s", CORRECT_PATH_SEPARATOR, pLocation );
-				FileSystem_AddLoadedSearchPath( initInfo, pPathID, &bFirstGamePath, baseDir, ep1EraPath, bLowViolence );
 		}
+
+		CUtlStringList vecFullLocationPaths;
+		char szAbsSearchPath[MAX_PATH];
+		V_MakeAbsolutePath( szAbsSearchPath, sizeof( szAbsSearchPath ), pLocation, pszBaseDir );
+
+		// Now resolve any ./'s.
+		V_FixSlashes( szAbsSearchPath );
+		if ( !V_RemoveDotSlashes( szAbsSearchPath ) )
+			Error( "FileSystem_AddLoadedSearchPath - Can't resolve pathname for '%s'", szAbsSearchPath );
+		V_StripTrailingSlash( szAbsSearchPath );
+
+		// Don't bother doing any wildcard expansion unless it has wildcards.  This avoids the weird
+		// thing with xxx_dir.vpk files being referred to simply as xxx.vpk.
+		if ( V_stristr( pLocation, "?") == NULL && V_stristr( pLocation, "*") == NULL )
+		{
+			vecFullLocationPaths.CopyAndAddToTail( szAbsSearchPath );
 		}
 		else
 		{
-			FileSystem_AddLoadedSearchPath( initInfo, pPathID, &bFirstGamePath, baseDir, pLocation, bLowViolence );
+			FileFindHandle_t findHandle = 0;
+			const char *pszFoundShortName = initInfo.m_pFileSystem->FindFirst( szAbsSearchPath, &findHandle );
+			if ( pszFoundShortName )
+			{
+				do 
+				{
+
+					// We only know how to mount VPK's and directories
+					if ( pszFoundShortName[0] != '.' && ( initInfo.m_pFileSystem->FindIsDirectory( findHandle ) || V_stristr( pszFoundShortName, ".vpk" ) ) )
+					{
+						char szAbsName[MAX_PATH];
+						V_ExtractFilePath( szAbsSearchPath, szAbsName, sizeof( szAbsName ) );
+						V_AppendSlash( szAbsName, sizeof(szAbsName) );
+						V_strcat_safe( szAbsName, pszFoundShortName );
+
+						vecFullLocationPaths.CopyAndAddToTail( szAbsName );
+
+						// Check for a common mistake
+						if (
+							!V_stricmp( pszFoundShortName, "materials" )
+							|| !V_stricmp( pszFoundShortName, "maps" )
+							|| !V_stricmp( pszFoundShortName, "resource" )
+							|| !V_stricmp( pszFoundShortName, "scripts" )
+							|| !V_stricmp( pszFoundShortName, "sound" )
+							|| !V_stricmp( pszFoundShortName, "models" ) )
+						{
+
+							char szReadme[MAX_PATH];
+							V_ExtractFilePath( szAbsSearchPath, szReadme, sizeof( szReadme ) );
+							V_AppendSlash( szReadme, sizeof(szReadme) );
+							V_strcat_safe( szReadme, "readme.txt" );
+
+							Error(
+								"Tried to add %s as a search path.\n"
+								"\nThis is probably not what you intended.\n"
+								"\nCheck %s for more info\n",
+								szAbsName, szReadme );
+						}
+
+					}
+					pszFoundShortName = initInfo.m_pFileSystem->FindNext( findHandle );
+				} while ( pszFoundShortName );
+				initInfo.m_pFileSystem->FindClose( findHandle );
+			}
+
+			// Sort alphabetically.  Also note that this will put
+			// all the xxx_000.vpk packs just before the corresponding
+			// xxx_dir.vpk
+			vecFullLocationPaths.Sort( SortStricmp );
+
+			// Now for any _dir.vpk files, remove the _nnn.vpk ones.
+			int idx = vecFullLocationPaths.Count()-1;
+			while ( idx > 0 )
+			{
+				char szTemp[ MAX_PATH ];
+				V_strcpy_safe( szTemp, vecFullLocationPaths[ idx ] );
+				--idx;
+
+				char *szDirVpk = V_stristr( szTemp, "_dir.vpk" );
+				if ( szDirVpk != NULL )
+				{
+					*szDirVpk = '\0';
+					while ( idx >= 0 )
+					{
+						char *pszPath = vecFullLocationPaths[ idx ];
+						if ( V_stristr( pszPath, szTemp ) != pszPath )
+							break;
+						delete pszPath;
+						vecFullLocationPaths.Remove( idx );
+						--idx;
+					}
+				}
+			}
+		}
+
+		// Parse Path ID list
+		CUtlStringList vecPathIDs;
+		V_SplitString( pCur->GetName(), "+", vecPathIDs );
+		FOR_EACH_VEC( vecPathIDs, idxPathID )
+		{
+			Q_StripPrecedingAndTrailingWhitespace( vecPathIDs[ idxPathID ] );
+		}
+
+		// Mount them.
+		FOR_EACH_VEC( vecFullLocationPaths, idxLocation )
+		{
+			FOR_EACH_VEC( vecPathIDs, idxPathID )
+			{
+				FileSystem_AddLoadedSearchPath( initInfo, vecPathIDs[ idxPathID ], vecFullLocationPaths[ idxLocation ], bLowViolence );
+			}
 		}
 	}
 
@@ -679,15 +709,13 @@ FSReturnCode_t FileSystem_LoadSearchPaths( CFSSearchPathsInit &initInfo )
 	// when people forget to specify a search path.
 	initInfo.m_pFileSystem->MarkPathIDByRequestOnly( "executable_path", true );
 	initInfo.m_pFileSystem->MarkPathIDByRequestOnly( "gamebin", true );
+	initInfo.m_pFileSystem->MarkPathIDByRequestOnly( "download", true );
 	initInfo.m_pFileSystem->MarkPathIDByRequestOnly( "mod", true );
-	if ( initInfo.m_ModPath[0] != 0 )
-	{
-		// Add the write path last.
-		initInfo.m_pFileSystem->AddSearchPath( initInfo.m_ModPath, "DEFAULT_WRITE_PATH", PATH_ADD_TO_TAIL );
-	}
+	initInfo.m_pFileSystem->MarkPathIDByRequestOnly( "game_write", true );
+	initInfo.m_pFileSystem->MarkPathIDByRequestOnly( "mod_write", true );
 
 #ifdef _DEBUG	
-	initInfo.m_pFileSystem->PrintSearchPaths();
+	// initInfo.m_pFileSystem->PrintSearchPaths();
 #endif
 
 	return FS_OK;
@@ -954,60 +982,6 @@ bool DoesPathExistAlready( const char *pPathEnvVar, const char *pTestPath )
 	}
 }
 
-FSReturnCode_t SetSteamInstallPath( char *steamInstallPath, int steamInstallPathLen, CSteamEnvVars &steamEnvVars, bool bErrorsAsWarnings )
-{
-	if ( IsConsole() )
-	{
-		// consoles don't use steam
-		return FS_MISSING_STEAM_DLL;
-	}
-
-	// Start at our bin directory and move up until we find a directory with steam.dll in it.
-	char executablePath[MAX_PATH];
-	if ( !FileSystem_GetExecutableDir( executablePath, sizeof( executablePath ) )	)
-	{
-		if ( bErrorsAsWarnings )
-		{
-			Warning( "SetSteamInstallPath: FileSystem_GetExecutableDir failed.\n" );
-			return FS_INVALID_PARAMETERS;
-		}
-		else
-		{
-			return SetupFileSystemError( false, FS_INVALID_PARAMETERS, "FileSystem_GetExecutableDir failed." );
-		}
-	}
-
-	Q_strncpy( steamInstallPath, executablePath, steamInstallPathLen );
-	while ( 1 )
-	{
-		// Ignore steamapp.cfg here in case they're debugging. We still need to know the real steam path so we can find their username.
-		// find 
-		if ( DoesFileExistIn( steamInstallPath, "steam.dll" ) && !DoesFileExistIn( steamInstallPath, "steamapp.cfg" ) )
-			break;
-	
-		if ( !Q_StripLastDir( steamInstallPath, steamInstallPathLen ) )
-		{
-			if ( bErrorsAsWarnings )
-			{
-				Warning( "Can't find steam.dll relative to executable path: %s.\n", executablePath );
-				return FS_MISSING_STEAM_DLL;
-			}
-			else
-			{
-				return SetupFileSystemError( false, FS_MISSING_STEAM_DLL, "Can't find steam.dll relative to executable path: %s.", executablePath );
-			}
-		}			
-	}
-
-	// Also, add the install path to their PATH environment variable, so filesystem_steam.dll can get to steam.dll.
-	char szPath[ 8192 ];
-	steamEnvVars.m_Path.GetValue( szPath, sizeof( szPath ) );
-	if ( !DoesPathExistAlready( szPath, steamInstallPath ) )
-	{
-		steamEnvVars.m_Path.SetValue( "%s;%s", szPath, steamInstallPath );
-	}
-	return FS_OK;
-}
 
 FSReturnCode_t GetSteamCfgPath( char *steamCfgPath, int steamCfgPathLen )
 {
@@ -1087,53 +1061,6 @@ void SetSteamUserPassphrase( KeyValues *pSteamInfo, CSteamEnvVars &steamEnvVars
 	}
 }
 
-void SetSteamAppId( KeyValues *pFileSystemInfo, const char *pGameInfoDirectory, CSteamEnvVars &steamEnvVars )
-{
-	// SteamAppId is in gameinfo.txt->FileSystem->FileSystemInfo_Steam->SteamAppId.
-	int iAppId = pFileSystemInfo->GetInt( "SteamAppId", -1 );
-	if ( iAppId == -1 )
-		Error( "Missing SteamAppId in %s\\%s.", pGameInfoDirectory, GAMEINFO_FILENAME );
-
-	steamEnvVars.m_SteamAppId.SetValue( "%d", iAppId );
-}
-
-FSReturnCode_t SetupSteamStartupEnvironment( KeyValues *pFileSystemInfo, const char *pGameInfoDirectory, CSteamEnvVars &steamEnvVars )
-{
-	// Ok, we're going to run Steam. See if they have SteamInfo.txt. If not, we'll try to deduce what we can.
-	char steamInfoFile[MAX_PATH];
-	Q_strncpy( steamInfoFile, pGameInfoDirectory, sizeof( steamInfoFile ) );
-	Q_AppendSlash( steamInfoFile, sizeof( steamInfoFile ) );
-	Q_strncat( steamInfoFile, "steaminfo.txt", sizeof( steamInfoFile ), COPY_ALL_CHARACTERS );
-	KeyValues *pSteamInfo = ReadKeyValuesFile( steamInfoFile );
-
-	char steamInstallPath[MAX_PATH];
-	FSReturnCode_t ret = SetSteamInstallPath( steamInstallPath, sizeof( steamInstallPath ), steamEnvVars, false );
-	if ( ret != FS_OK )
-		return ret;
-
-	SetSteamAppUser( pSteamInfo, steamInstallPath, steamEnvVars );
-	SetSteamUserPassphrase( pSteamInfo, steamEnvVars );
-	SetSteamAppId( pFileSystemInfo, pGameInfoDirectory, steamEnvVars );
-
-	if ( pSteamInfo )
-		pSteamInfo->deleteThis();
-
-	return FS_OK;
-}
-
-FSReturnCode_t GetSteamExtraAppId( const char *pDirectoryName, int *nExtraAppId )
-{
-	// Now, load gameinfo.txt (to make sure it's there)
-	KeyValues *pMainFile, *pFileSystemInfo, *pSearchPaths;
-	FSReturnCode_t ret = LoadGameInfoFile( pDirectoryName, pMainFile, pFileSystemInfo, pSearchPaths );
-	if ( ret != FS_OK )
-		return ret;
-
-	*nExtraAppId = pFileSystemInfo->GetInt( "ToolsAppId", -1 );
-	pMainFile->deleteThis();
-	return FS_OK;
-}
-
 FSReturnCode_t FileSystem_SetBasePaths( IFileSystem *pFileSystem )
 {
 	pFileSystem->RemoveSearchPaths( "EXECUTABLE_PATH" );
@@ -1143,6 +1070,12 @@ FSReturnCode_t FileSystem_SetBasePaths( IFileSystem *pFileSystem )
 		return SetupFileSystemError( false, FS_INVALID_PARAMETERS, "FileSystem_GetExecutableDir failed." );
 
 	pFileSystem->AddSearchPath( executablePath, "EXECUTABLE_PATH" );
+
+	if ( !FileSystem_GetBaseDir( executablePath, sizeof( executablePath ) )  )
+		return SetupFileSystemError( false, FS_INVALID_PARAMETERS, "FileSystem_GetBaseDir failed." );
+
+	pFileSystem->AddSearchPath( executablePath, "BASE_PATH" );
+
 	return FS_OK;
 }
 
@@ -1159,40 +1092,30 @@ FSReturnCode_t FileSystem_GetFileSystemDLLName( char *pFileSystemDLL, int nMaxLe
 	if ( !FileSystem_GetExecutableDir( executablePath, sizeof( executablePath ) )	)
 		return SetupFileSystemError( false, FS_INVALID_PARAMETERS, "FileSystem_GetExecutableDir failed." );
 
-#if defined( _WIN32 ) && !defined( _X360 )
-	// If filesystem_stdio.dll is missing or -steam is specified, then load filesystem_steam.dll.
-	// There are two command line parameters for Steam:
-	//		1) -steam (runs Steam in remote filesystem mode; requires Steam backend)
-	//		2) -steamlocal (runs Steam in local filesystem mode (all content off HDD)
-	Q_snprintf( pFileSystemDLL, nMaxLen, "%s%cfilesystem_stdio.dll", executablePath, CORRECT_PATH_SEPARATOR );
-	if ( CommandLine()->FindParm( "-steam" ) || CommandLine()->FindParm( "-steamlocal" ) || _access( pFileSystemDLL, 0 ) != 0 )
-	{
-		Q_snprintf( pFileSystemDLL, nMaxLen, "%s%cfilesystem_steam.dll", executablePath, CORRECT_PATH_SEPARATOR );
+	// Assume we'll use local files
+	Q_snprintf( pFileSystemDLL, nMaxLen, "%s%cfilesystem_stdio" DLL_EXT_STRING, executablePath, CORRECT_PATH_SEPARATOR );
+
+	#if !defined( _X360 )
+
+		// Use filsystem_steam if it exists?
+		#if defined( OSX ) || defined( LINUX )
+			struct stat statBuf;
+		#endif
+		if (
+			#if defined( OSX ) || defined( LINUX )
+				stat( pFileSystemDLL, &statBuf ) != 0
+			#else
+				_access( pFileSystemDLL, 0 ) != 0
+			#endif
+		) {
+			Q_snprintf( pFileSystemDLL, nMaxLen, "%s%cfilesystem_steam" DLL_EXT_STRING, executablePath, CORRECT_PATH_SEPARATOR );
 			bSteam = true;
 		}
-#elif defined( _X360 )
-	Q_snprintf( pFileSystemDLL, nMaxLen, "%s%cfilesystem_stdio.dll", executablePath, CORRECT_PATH_SEPARATOR );
-#elif defined( _LINUX )
-	Q_snprintf( pFileSystemDLL, nMaxLen, "%s%cfilesystem_i486.so", executablePath, CORRECT_PATH_SEPARATOR );
-#else
-	#error "define a filesystem dll name"
 	#endif
 
 	return FS_OK;
 }
 
-//-----------------------------------------------------------------------------
-// Sets up the steam.dll install path in our PATH env var (so you can then just 
-// LoadLibrary() on filesystem_steam.dll without having to copy steam.dll anywhere special )
-//-----------------------------------------------------------------------------
-FSReturnCode_t FileSystem_SetupSteamInstallPath()
-{
-	CSteamEnvVars steamEnvVars;
-	char steamInstallPath[MAX_PATH];
-	FSReturnCode_t ret = SetSteamInstallPath( steamInstallPath, sizeof( steamInstallPath ), steamEnvVars, true );
-	steamEnvVars.m_Path.SetRestoreOriginalValue( false ); // We want to keep the change to the path going forward.
-	return ret;
-}
 
 //-----------------------------------------------------------------------------
 // Sets up the steam environment + gets back the gameinfo.txt path
@@ -1205,45 +1128,13 @@ FSReturnCode_t FileSystem_SetupSteamEnvironment( CFSSteamSetupInfo &fsInfo )
 		return ret;
 
 	// This is so that processes spawned by this application will have the same VPROJECT
+#ifdef WIN32
 	char pEnvBuf[MAX_PATH+32];
 	Q_snprintf( pEnvBuf, sizeof(pEnvBuf), "%s=%s", GAMEDIR_TOKEN, fsInfo.m_GameInfoPath );
 	_putenv( pEnvBuf );
-
-	CSteamEnvVars steamEnvVars;
-	if ( fsInfo.m_bSteam )
-	{
-		if ( fsInfo.m_bToolsMode )
-		{
-			// Now, load gameinfo.txt (to make sure it's there)
-			KeyValues *pMainFile, *pFileSystemInfo, *pSearchPaths;
-			ret = LoadGameInfoFile( fsInfo.m_GameInfoPath, pMainFile, pFileSystemInfo, pSearchPaths );
-			if ( ret != FS_OK )
-				return ret;
-
-			// If filesystem_stdio.dll is missing or -steam is specified, then load filesystem_steam.dll.
-			// There are two command line parameters for Steam:
-			//		1) -steam (runs Steam in remote filesystem mode; requires Steam backend)
-			//		2) -steamlocal (runs Steam in local filesystem mode (all content off HDD)
-
-			// Setup all the environment variables related to Steam so filesystem_steam.dll knows how to initialize Steam.
-			ret = SetupSteamStartupEnvironment( pFileSystemInfo, fsInfo.m_GameInfoPath, steamEnvVars );
-			if ( ret != FS_OK )
-				return ret;
-
-			steamEnvVars.m_SteamAppId.SetRestoreOriginalValue( false ); // We want to keep the change to the path going forward.
-
-			// We're done with main file
-			pMainFile->deleteThis();
-		}
-		else if ( fsInfo.m_bSetSteamDLLPath )
-		{
-			// This is used by the engine to automatically set the path to their steam.dll when running the engine,
-			// so they can debug it without having to copy steam.dll up into their hl2.exe folder.
-			char steamInstallPath[MAX_PATH];
-			ret = SetSteamInstallPath( steamInstallPath, sizeof( steamInstallPath ), steamEnvVars, true );
-			steamEnvVars.m_Path.SetRestoreOriginalValue( false ); // We want to keep the change to the path going forward.
-		}
-	}
+#else
+	setenv( GAMEDIR_TOKEN, fsInfo.m_GameInfoPath, 1 );
+#endif
 
 	return FS_OK;
 }
@@ -1287,33 +1178,33 @@ FSReturnCode_t FileSystem_MountContent( CFSMountContentInfo &mountContentInfo )
 	// This part is Steam-only.
 	if ( mountContentInfo.m_pFileSystem->IsSteam() )
 	{
-		// Find out the "extra app id". This is for tools, which want to mount a base app's filesystem
-		// like HL2, then mount the SDK content (tools materials and models, etc) in addition.
-		int nExtraAppId = -1;
-		if ( mountContentInfo.m_bToolsMode )
-		{
-			FSReturnCode_t ret = GetSteamExtraAppId( mountContentInfo.m_pDirectoryName, &nExtraAppId );
-			if ( ret != FS_OK )
-				return ret;
-		}
+		return SetupFileSystemError( false, FS_INVALID_PARAMETERS, "Should not be using filesystem_steam anymore!" );
 
-		// Set our working directory temporarily so Steam can remember it.
-		// This is what Steam strips off absolute filenames like c:\program files\valve\steam\steamapps\username\sourcesdk
-		// to get to the relative part of the path.
-		char baseDir[MAX_PATH], oldWorkingDir[MAX_PATH];
-		if ( !FileSystem_GetBaseDir( baseDir, sizeof( baseDir ) ) )
-			return SetupFileSystemError( false, FS_INVALID_PARAMETERS, "FileSystem_GetBaseDir failed." );
-
-		Q_getwd( oldWorkingDir, sizeof( oldWorkingDir ) );
-		_chdir( baseDir );
-
-		// Filesystem_tools needs to add dependencies in here beforehand.
-		FilesystemMountRetval_t retVal = mountContentInfo.m_pFileSystem->MountSteamContent( nExtraAppId );
-		
-		_chdir( oldWorkingDir );
-
-		if ( retVal != FILESYSTEM_MOUNT_OK )
-			return SetupFileSystemError( true, FS_UNABLE_TO_INIT, "Unable to mount Steam content in the file system" );
+//		// Find out the "extra app id". This is for tools, which want to mount a base app's filesystem
+//		// like HL2, then mount the SDK content (tools materials and models, etc) in addition.
+//		int nExtraAppId = -1;
+//		if ( mountContentInfo.m_bToolsMode )
+//		{
+//			// !FIXME! Here we need to mount the tools content (VPK's) in some way...?
+//		}
+//
+//		// Set our working directory temporarily so Steam can remember it.
+//		// This is what Steam strips off absolute filenames like c:\program files\valve\steam\steamapps\username\sourcesdk
+//		// to get to the relative part of the path.
+//		char baseDir[MAX_PATH], oldWorkingDir[MAX_PATH];
+//		if ( !FileSystem_GetBaseDir( baseDir, sizeof( baseDir ) ) )
+//			return SetupFileSystemError( false, FS_INVALID_PARAMETERS, "FileSystem_GetBaseDir failed." );
+//
+//		Q_getwd( oldWorkingDir, sizeof( oldWorkingDir ) );
+//		_chdir( baseDir );
+//
+//		// Filesystem_tools needs to add dependencies in here beforehand.
+//		FilesystemMountRetval_t retVal = mountContentInfo.m_pFileSystem->MountSteamContent( nExtraAppId );
+//		
+//		_chdir( oldWorkingDir );
+//
+//		if ( retVal != FILESYSTEM_MOUNT_OK )
+//			return SetupFileSystemError( true, FS_UNABLE_TO_INIT, "Unable to mount Steam content in the file system" );
 	}
 
 	return FileSystem_SetBasePaths( mountContentInfo.m_pFileSystem );
@@ -1342,17 +1233,9 @@ void FileSystem_ClearSteamEnvVars()
 void FileSystem_AddSearchPath_Platform( IFileSystem *pFileSystem, const char *szGameInfoPath )
 {
 	char platform[MAX_PATH];
-	if ( pFileSystem->IsSteam() )
-	{
-		// Steam doesn't support relative paths
-		Q_strncpy( platform, "platform", MAX_PATH );
-	}
-	else
-	{
 	Q_strncpy( platform, szGameInfoPath, MAX_PATH );
 	Q_StripTrailingSlash( platform );
 	Q_strncat( platform, "/../platform", MAX_PATH, MAX_PATH );
-	}
 
 	pFileSystem->AddSearchPath( platform, "PLATFORM" );
 }

public/filesystem_init.h

@@ -1,4 +1,4 @@
-//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
+//========= Copyright Valve Corporation, All rights reserved. ============//
 //
 // Purpose: 
 //
@@ -148,14 +148,17 @@ public:
 	// This specifies the directory where gameinfo.txt is. This must be set.
 	const char		*m_pDirectoryName;
 
-	// If this is set, then it will add a search path with _<language> appended to the pathname
-	// for each search path with a path ID of "game".
+	// If this is set, then any search paths with a _english will be replaced with _m_pLanguage and added before the
+	// _english path
 	// (default: null)
 	const char		*m_pLanguage;
 
 	// This is the filesystem FileSystem_LoadSearchPaths is talking to.
 	IFileSystem		*m_pFileSystem;
 
+	bool m_bMountHDContent;
+	bool m_bLowViolence;
+
 // Outputs.
 public:
 	// This is the location of the first search path called "game", which also becomes your "mod" search path.
@@ -206,11 +209,6 @@ void FileSystem_ClearSteamEnvVars();
 // Find the steam.cfg above you for optional stuff
 FSReturnCode_t GetSteamCfgPath( char *steamCfgPath, int steamCfgPathLen );
 
-// Setup the Steam.dll path without needing all the extra gameinfo stuff first
-// used by the CSteamApplication::Create() code to LoadModule() on the filesystem
-// before the underlying apps know specific details about the environment to load
-FSReturnCode_t FileSystem_SetupSteamInstallPath();
-
 // Returns the last error.
 const char *FileSystem_GetLastErrorString();
 

public/filesystem_passthru.h

@@ -52,7 +52,7 @@ public:
 	virtual bool			FileExists( const char *pFileName, const char *pPathID )					{ return m_pBaseFileSystemPassThru->FileExists( pFileName, pPathID ); }
 	virtual bool			IsFileWritable( char const *pFileName, const char *pPathID )					{ return m_pBaseFileSystemPassThru->IsFileWritable( pFileName, pPathID ); }
 	virtual bool			SetFileWritable( char const *pFileName, bool writable, const char *pPathID )	{ return m_pBaseFileSystemPassThru->SetFileWritable( pFileName, writable, pPathID ); }
-	virtual long			GetFileTime( const char *pFileName, const char *pPathID )						{ return m_pBaseFileSystemPassThru->GetFileTime( pFileName, pPathID ); }
+	virtual int32_t			GetFileTime( const char *pFileName, const char *pPathID )						{ return m_pBaseFileSystemPassThru->GetFileTime( pFileName, pPathID ); }
 	virtual bool			ReadFile( const char *pFileName, const char *pPath, CUtlBuffer &buf, int nMaxBytes = 0, int nStartingByte = 0, FSAllocFunc_t pfnAlloc = NULL ) { return m_pBaseFileSystemPassThru->ReadFile( pFileName, pPath, buf, nMaxBytes, nStartingByte, pfnAlloc  ); }
 	virtual bool			WriteFile( const char *pFileName, const char *pPath, CUtlBuffer &buf )		{  return m_pBaseFileSystemPassThru->WriteFile( pFileName, pPath, buf ); }
 	virtual bool			UnzipFile( const char *pFileName, const char *pPath, const char *pDestination )		{  return m_pBaseFileSystemPassThru->UnzipFile( pFileName, pPath, pDestination ); }
@@ -100,7 +100,7 @@ public:
 	virtual bool			RenameFile( char const *pOldPath, char const *pNewPath, const char *pathID )		{ return m_pFileSystemPassThru->RenameFile( pOldPath, pNewPath, pathID ); }
 	virtual void			CreateDirHierarchy( const char *path, const char *pathID )							{ m_pFileSystemPassThru->CreateDirHierarchy( path, pathID ); }
 	virtual bool			IsDirectory( const char *pFileName, const char *pathID )							{ return m_pFileSystemPassThru->IsDirectory( pFileName, pathID ); }
-	virtual void			FileTimeToString( char* pStrip, int maxCharsIncludingTerminator, long fileTime )	{ m_pFileSystemPassThru->FileTimeToString( pStrip, maxCharsIncludingTerminator, fileTime ); }
+	virtual void			FileTimeToString( char* pStrip, int maxCharsIncludingTerminator, int32_t fileTime )	{ m_pFileSystemPassThru->FileTimeToString( pStrip, maxCharsIncludingTerminator, fileTime ); }
 	virtual void			SetBufferSize( FileHandle_t file, unsigned nBytes )									{ m_pFileSystemPassThru->SetBufferSize( file, nBytes  ); }
 	virtual bool			IsOk( FileHandle_t file )															{ return m_pFileSystemPassThru->IsOk( file ); }
 	virtual bool			EndOfFile( FileHandle_t file )														{ return m_pFileSystemPassThru->EndOfFile( file ); }
@@ -211,7 +211,7 @@ public:
 	virtual bool			ReadToBuffer( FileHandle_t hFile, CUtlBuffer &buf, int nMaxBytes = 0, FSAllocFunc_t pfnAlloc = NULL ) { return m_pFileSystemPassThru->ReadToBuffer( hFile, buf, nMaxBytes, pfnAlloc ); }
 	virtual bool			FullPathToRelativePathEx( const char *pFullPath, const char *pPathId, char *pRelative, int nMaxLen ) { return m_pFileSystemPassThru->FullPathToRelativePathEx( pFullPath, pPathId, pRelative, nMaxLen ); }
 	virtual int				GetPathIndex( const FileNameHandle_t &handle ) { return m_pFileSystemPassThru->GetPathIndex( handle ); }
-	virtual long			GetPathTime( const char *pPath, const char *pPathID ) { return m_pFileSystemPassThru->GetPathTime( pPath, pPathID ); }
+	virtual int32_t			GetPathTime( const char *pPath, const char *pPathID ) { return m_pFileSystemPassThru->GetPathTime( pPath, pPathID ); }
 
 	virtual DVDMode_t		GetDVDMode() { return m_pFileSystemPassThru->GetDVDMode(); }
 

public/icvar.h

@@ -102,8 +102,91 @@ public:
 #if defined( _X360 )
 	virtual void			PublishToVXConsole( ) = 0;
 #endif
+	virtual bool			IsMaterialThreadSetAllowed( ) const = 0;
+	virtual void			QueueMaterialThreadSetValue( ConVar *pConVar, const char *pValue ) = 0;
+	virtual void			QueueMaterialThreadSetValue( ConVar *pConVar, int nValue ) = 0;
+	virtual void			QueueMaterialThreadSetValue( ConVar *pConVar, float flValue ) = 0;
+	virtual bool			HasQueuedMaterialThreadConVarSets() const = 0;
+	virtual int				ProcessQueuedMaterialThreadConVarSets() = 0;
+
+protected:	class ICVarIteratorInternal;
+public:
+	/// Iteration over all cvars. 
+	/// (THIS IS A SLOW OPERATION AND YOU SHOULD AVOID IT.)
+	/// usage: 
+	/// { ICVar::Iterator iter(g_pCVar); 
+	///   for ( iter.SetFirst() ; iter.IsValid() ; iter.Next() )
+	///   {  
+	///       ConCommandBase *cmd = iter.Get();
+	///   } 
+	/// }
+	/// The Iterator class actually wraps the internal factory methods
+	/// so you don't need to worry about new/delete -- scope takes care
+	//  of it.
+	/// We need an iterator like this because we can't simply return a 
+	/// pointer to the internal data type that contains the cvars -- 
+	/// it's a custom, protected class with unusual semantics and is
+	/// prone to change.
+	class Iterator
+	{
+	public:
+		inline Iterator(ICvar *icvar);
+		inline ~Iterator(void);
+		inline void		SetFirst( void );
+		inline void		Next( void );
+		inline bool		IsValid( void );
+		inline ConCommandBase *Get( void );
+	private:
+		ICVarIteratorInternal *m_pIter;
 	};
 
+protected:
+	// internals for  ICVarIterator
+	class ICVarIteratorInternal
+	{
+	public:
+		// warning: delete called on 'ICvar::ICVarIteratorInternal' that is abstract but has non-virtual destructor [-Wdelete-non-virtual-dtor]
+		virtual ~ICVarIteratorInternal() {}
+		virtual void		SetFirst( void ) = 0;
+		virtual void		Next( void ) = 0;
+		virtual	bool		IsValid( void ) = 0;
+		virtual ConCommandBase *Get( void ) = 0;
+	};
+
+	virtual ICVarIteratorInternal	*FactoryInternalIterator( void ) = 0;
+	friend class Iterator;
+};
+
+inline ICvar::Iterator::Iterator(ICvar *icvar)
+{
+	m_pIter = icvar->FactoryInternalIterator();
+}
+
+inline ICvar::Iterator::~Iterator( void )
+{
+	delete m_pIter;
+}
+
+inline void ICvar::Iterator::SetFirst( void )
+{
+	m_pIter->SetFirst();
+}
+
+inline void ICvar::Iterator::Next( void )
+{
+	m_pIter->Next();
+}
+
+inline bool ICvar::Iterator::IsValid( void )
+{
+	return m_pIter->IsValid();
+}
+
+inline ConCommandBase * ICvar::Iterator::Get( void )
+{
+	return m_pIter->Get();
+}
+
 #define CVAR_INTERFACE_VERSION "VEngineCvar004"
 
 
@@ -114,7 +197,7 @@ public:
 
 // These are marked DLL_EXPORT for Linux.
 DLL_EXPORT ICvar *cvar;
-DLL_EXPORT ICvar *g_pCVar;
+extern ICvar *g_pCVar;
 
 
 #endif // ICVAR_H

public/inetchannel.h

@@ -28,8 +28,6 @@ public:
 
 	virtual void	SetDataRate(float rate) = 0;
 	virtual bool	RegisterMessage(INetMessage *msg) = 0;
-	virtual bool	StartStreaming( unsigned int challengeNr ) = 0;
-	virtual void	ResetStreaming( void ) = 0;
 	virtual void	SetTimeout(float seconds) = 0;
 	virtual void	SetDemoRecorder(IDemoRecorder *recorder) = 0;
 	virtual void	SetChallengeNr(unsigned int chnr) = 0;
@@ -39,7 +37,6 @@ public:
 	virtual void	Shutdown(const char *reason) = 0;
 	
 	virtual void	ProcessPlayback( void ) = 0;
-	virtual bool	ProcessStream( void ) = 0;
 	virtual void	ProcessPacket( struct netpacket_s* packet, bool bHasHeader ) = 0;
 			
 	virtual bool	SendNetMsg(INetMessage &msg, bool bForceReliable = false, bool bVoice = false ) = 0;
@@ -72,8 +69,6 @@ public:
 	virtual void	SetCompressionMode( bool bUseCompression ) = 0;
 	virtual unsigned int RequestFile(const char *filename) = 0;
 	
-	virtual float	GetTimeSinceLastReceived( void ) const = 0;
-
 	virtual void	SetMaxBufferSize(bool bReliable, int nBytes, bool bVoice = false ) = 0;
 
 	virtual bool	IsNull() const = 0;

public/iserver.h

@@ -1,4 +1,4 @@
-//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
+//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
 //
 // Purpose: 
 //
@@ -13,6 +13,7 @@
 #include <inetmsghandler.h>
 #include <bitvec.h>
 #include <const.h>
+#include <netadr.h>
 
 class INetMessage;
 class IRecipientFilter;
@@ -31,7 +32,9 @@ public:
 	virtual int		GetMaxClients( void ) const = 0; // returns current client limit
 	virtual IClient	*GetClient( int index ) = 0; // returns interface to client 
 	virtual int		GetClientCount() const = 0; // returns number of clients slots (used & unused)
-	virtual int		GetUDPPort( void ) const = 0; // returns current used UDP port
+	virtual netadr_t GetPublicAddress() = 0;
+	virtual bool		IsUsingFakeIP( void ) const = 0;
+	virtual int		GetLocalUDPPort( void ) const = 0; // returns current used UDP port
 	virtual float	GetTime( void ) const = 0;	// returns game world time
 	virtual int		GetTick( void ) const = 0;	// returns game world tick
 	virtual float	GetTickInterval( void ) const = 0; // tick interval in seconds

public/ivrenderview.h

@@ -209,7 +209,7 @@ public:
 	virtual IWorldRenderList * CreateWorldList() = 0;
 
 	virtual void			BuildWorldLists( IWorldRenderList *pList, WorldListInfo_t* pInfo, int iForceFViewLeaf, const VisOverrideData_t* pVisData = NULL, bool bShadowDepth = false, float *pReflectionWaterHeight = NULL ) = 0;
-	virtual void			DrawWorldLists( IWorldRenderList *pList, unsigned long flags, float waterZAdjust ) = 0;
+	virtual void			DrawWorldLists( IWorldRenderList *pList, uint32_t flags, float waterZAdjust ) = 0;
 
 	// Optimization for top view
 	virtual void			DrawTopView( bool enable ) = 0;
@@ -221,7 +221,7 @@ public:
 	virtual void			DrawMaskEntities( void ) = 0;
 
 	// Draw surfaces with alpha
-	virtual void			DrawTranslucentSurfaces( IWorldRenderList *pList, int sortIndex, unsigned long flags, bool bShadowDepth ) = 0;
+	virtual void			DrawTranslucentSurfaces( IWorldRenderList *pList, int sortIndex, uint32_t flags, bool bShadowDepth ) = 0;
 
 	// Draw Particles ( just draws the linefine for debugging map leaks )
 	virtual void			DrawLineFile( void ) = 0;
@@ -257,7 +257,7 @@ public:
 	virtual void			DrawBrushModelShadow( IClientRenderable *pRenderable ) = 0;
 
 	// Does the leaf contain translucent surfaces?
-	virtual	bool			LeafContainsTranslucentSurfaces( IWorldRenderList *pList, int sortIndex, unsigned long flags ) = 0;
+	virtual	bool			LeafContainsTranslucentSurfaces( IWorldRenderList *pList, int sortIndex, uint32_t flags ) = 0;
 
 	virtual bool			DoesBoxIntersectWaterVolume( const Vector &mins, const Vector &maxs, int leafWaterDataID ) = 0;
 

public/materialsystem/imaterialsystem.h

@@ -43,6 +43,7 @@ struct MaterialSystem_Config_t;
 class VMatrix;
 struct matrix3x4_t;
 class ITexture;
+class ITextureCompositor;
 struct MaterialSystemHardwareIdentifier_t;
 class KeyValues;
 class IShader;
@@ -65,7 +66,13 @@ typedef uint64 VertexFormat_t;
 
 // NOTE NOTE NOTE!!!!  If you up this, grep for "NEW_INTERFACE" to see if there is anything
 // waiting to be enabled during an interface revision.
-#define MATERIAL_SYSTEM_INTERFACE_VERSION "VMaterialSystem079"
+#define MATERIAL_SYSTEM_INTERFACE_VERSION "VMaterialSystem081"
+
+#ifdef POSIX
+#define ABSOLUTE_MINIMUM_DXLEVEL 90
+#else
+#define ABSOLUTE_MINIMUM_DXLEVEL 80
+#endif
 
 enum ShaderParamType_t 
 { 
@@ -82,6 +89,7 @@ enum ShaderParamType_t
 	SHADER_PARAM_TYPE_MATRIX,
 	SHADER_PARAM_TYPE_MATERIAL,
 	SHADER_PARAM_TYPE_STRING,
+	SHADER_PARAM_TYPE_MATRIX4X2
 };
 
 enum MaterialMatrixMode_t
@@ -217,6 +225,15 @@ enum MaterialContextType_t
 	MATERIAL_NULL_CONTEXT
 };
 
+//-----------------------------------------------------------------------------
+//
+//-----------------------------------------------------------------------------
+
+enum MaterialFindContext_t
+{
+	MATERIAL_FINDCONTEXT_NONE,
+	MATERIAL_FINDCONTEXT_ISONAMODEL,
+};
 
 //-----------------------------------------------------------------------------
 // Light structure
@@ -274,7 +291,6 @@ private:
 #define CREATERENDERTARGETFLAGS_NOEDRAM			0x00000008 // inhibit allocation in 360 EDRAM
 #define CREATERENDERTARGETFLAGS_TEMP			0x00000010 // only allocates memory upon first resolve, destroyed at level end
 
-
 //-----------------------------------------------------------------------------
 // allowed stencil operations. These match the d3d operations
 //-----------------------------------------------------------------------------
@@ -463,6 +479,21 @@ private:
 	int m_nBottom;
 };
 
+// Passed as the callback object to Async functions in the material system
+// so that callers don't have to worry about memory going out of scope before the 
+// results return.
+abstract_class IAsyncTextureOperationReceiver : public IRefCounted
+{
+public:
+	virtual void OnAsyncCreateComplete( ITexture* pTex, void* pExtraArgs ) = 0;
+	virtual void OnAsyncFindComplete( ITexture* pTex, void* pExtraArgs ) = 0;
+	virtual void OnAsyncMapComplete( ITexture* pTex, void* pExtraArgs, void* pMemory, int nPitch ) = 0;
+	virtual void OnAsyncReadbackBegin( ITexture* pDst, ITexture* pSrc, void* pExtraArgs ) = 0;
+
+	virtual int GetRefCount() const = 0;
+};
+
+
 //-----------------------------------------------------------------------------
 // Flags to be used with the Init call
 //-----------------------------------------------------------------------------
@@ -508,7 +539,11 @@ enum RenderTargetSizeMode_t
 	RT_SIZE_HDR=3,					// frame_buffer_width / 4
 	RT_SIZE_FULL_FRAME_BUFFER=4,	// Same size as frame buffer, or next lower power of 2 if we can't do that.
 	RT_SIZE_OFFSCREEN=5,			// Target of specified size, don't mess with dimensions
-	RT_SIZE_FULL_FRAME_BUFFER_ROUNDED_UP=6 // Same size as the frame buffer, rounded up if necessary for systems that can't do non-power of two textures.
+	RT_SIZE_FULL_FRAME_BUFFER_ROUNDED_UP=6, // Same size as the frame buffer, rounded up if necessary for systems that can't do non-power of two textures.
+	RT_SIZE_REPLAY_SCREENSHOT = 7,	// Rounded down to power of 2, essentially...
+	RT_SIZE_LITERAL = 8,			// Use the size passed in. Don't clamp it to the frame buffer size. Really.
+	RT_SIZE_LITERAL_PICMIP = 9		// Use the size passed in, don't clamp to the frame buffer size, but do apply picmip restrictions.
+
 };
 
 typedef void (*MaterialBufferReleaseFunc_t)( );
@@ -571,6 +606,7 @@ public:
 	//---------------------------------------------------------
 	virtual void					SetThreadMode( MaterialThreadMode_t mode, int nServiceThread = -1 ) = 0;
 	virtual MaterialThreadMode_t	GetThreadMode( ) = 0;
+	virtual bool					IsRenderThreadSafe( ) = 0;
 	virtual void					ExecuteQueued() = 0;
 
 	//---------------------------------------------------------
@@ -765,6 +801,8 @@ public:
 	//---------------------------------------------------------
 	// Material and texture management
 	//---------------------------------------------------------
+	virtual void				SuspendTextureStreaming( void ) = 0;
+	virtual void				ResumeTextureStreaming( void ) = 0;
 
 	// uncache all materials. .  good for forcing reload of materials.
 	virtual void				UncacheAllMaterials( ) = 0;
@@ -800,6 +838,9 @@ public:
 	// (Or use the global IsErrorMaterial function, which checks if it's null too).
 	virtual IMaterial *			FindMaterial( char const* pMaterialName, const char *pTextureGroupName, bool complain = true, const char *pComplainPrefix = NULL ) = 0;
 
+	// Query whether a material is loaded (eg, whether FindMaterial will be nonblocking)
+	virtual bool				IsMaterialLoaded( char const* pMaterialName ) = 0;
+
 	//---------------------------------
 	// This is the interface for knowing what materials are available
 	// is to use the following functions to get a list of materials.  The
@@ -827,7 +868,9 @@ public:
 
 	//---------------------------------
 
-	virtual ITexture *			FindTexture( char const* pTextureName, const char *pTextureGroupName, bool complain = true ) = 0;
+	virtual void				SetAsyncTextureLoadCache( void* hFileCache ) = 0;
+
+	virtual ITexture *			FindTexture( char const* pTextureName, const char *pTextureGroupName, bool complain = true, int nAdditionalCreationFlags = 0  ) = 0;
 
 	// Checks to see if a particular texture is loaded
 	virtual bool				IsTextureLoaded( char const* pTextureName ) const = 0;
@@ -997,6 +1040,49 @@ public:
 
 	// For sv_pure mode. The filesystem figures out which files the client needs to reload to be "pure" ala the server's preferences.
 	virtual void ReloadFilesInList( IFileList *pFilesToReload ) = 0;
+	virtual	bool				AllowThreading( bool bAllow, int nServiceThread ) = 0;
+
+	// Extended version of FindMaterial().
+	// Contains context in so it can make decisions (i.e. if it's a model, ignore certain cheat parameters)
+	virtual IMaterial *			FindMaterialEx( char const* pMaterialName, const char *pTextureGroupName, int nContext, bool complain = true, const char *pComplainPrefix = NULL ) = 0;
+
+#ifdef DX_TO_GL_ABSTRACTION
+	virtual void				DoStartupShaderPreloading( void ) = 0;
+#endif	
+
+	// Sets the override sizes for all render target size tests. These replace the frame buffer size.
+	// Set them when you are rendering primarily to something larger than the frame buffer (as in VR mode).
+	virtual void				SetRenderTargetFrameBufferSizeOverrides( int nWidth, int nHeight ) = 0;
+
+	// Returns the (possibly overridden) framebuffer size for render target sizing.
+	virtual void				GetRenderTargetFrameBufferDimensions( int & nWidth, int & nHeight ) = 0;
+
+	// returns the display device name that matches the adapter index we were started with
+	virtual char *GetDisplayDeviceName() const = 0;
+
+	// creates a texture suitable for use with materials from a raw stream of bits.
+	// The bits will be retained by the material system and can be freed upon return.
+	virtual ITexture*			CreateTextureFromBits(int w, int h, int mips, ImageFormat fmt, int srcBufferSize, byte* srcBits) = 0;
+
+	// Lie to the material system to pretend to be in render target allocation mode at the beginning of time.
+	// This was a thing that mattered a lot to old hardware, but doesn't matter at all to new hardware,
+	// where new is defined to be "anything from the last decade." However, we want to preserve legacy behavior
+	// for the old games because it's easier than testing them.
+	virtual void				OverrideRenderTargetAllocation( bool rtAlloc ) = 0;
+
+	// creates a texture compositor that will attempt to composite a new textuer from the steps of the specified KeyValues.
+	virtual ITextureCompositor*	NewTextureCompositor( int w, int h, const char* pCompositeName, int nTeamNum, uint64 randomSeed, KeyValues* stageDesc, uint32 texCompositeCreateFlags = 0 ) = 0;
+
+	// Loads the texture with the specified name, calls pRecipient->OnAsyncFindComplete with the result from the main thread.
+	// once the texture load is complete. If the texture cannot be found, the returned texture will return true for IsError().
+	virtual void AsyncFindTexture( const char* pFilename, const char *pTextureGroupName, IAsyncTextureOperationReceiver* pRecipient, void* pExtraArgs, bool bComplain = true, int nAdditionalCreationFlags = 0 ) = 0;
+
+	// creates a texture suitable for use with materials from a raw stream of bits.
+	// The bits will be retained by the material system and can be freed upon return.
+	virtual ITexture*			CreateNamedTextureFromBitsEx( const char* pName, const char *pTextureGroupName, int w, int h, int mips, ImageFormat fmt, int srcBufferSize, byte* srcBits, int nFlags ) = 0;
+	
+	virtual bool				AddTextureCompositorTemplate( const char *pName, KeyValues *kv, int a ) = 0;
+	virtual bool				VerifyTextureCompositorTemplates ( void ) = 0;
 };
 
 
@@ -1282,6 +1368,7 @@ public:
 
 	// Blit a subrect of the current render target to another texture
 	virtual void CopyRenderTargetToTextureEx( ITexture *pTexture, int nRenderTargetID, Rect_t *pSrcRect, Rect_t *pDstRect = NULL ) = 0;
+	virtual void CopyTextureToRenderTargetEx( int nRenderTargetID, ITexture *pTexture, Rect_t *pSrcRect, Rect_t *pDstRect = NULL ) = 0;
 
 	// Special off-center perspective matrix for DoF, MSAA jitter and poster rendering
 	virtual void PerspectiveOffCenterX( double fovx, double aspect, double zNear, double zFar, double bottom, double top, double left, double right ) = 0;
@@ -1341,9 +1428,9 @@ public:
 	virtual void GetFogDistances( float *fStart, float *fEnd, float *fFogZ ) = 0;
 
 	// Hooks for firing PIX events from outside the Material System...
-	virtual void BeginPIXEvent( unsigned long color, const char *szName ) = 0;
+	virtual void BeginPIXEvent( uint32_t color, const char *szName ) = 0;
 	virtual void EndPIXEvent() = 0;
-	virtual void SetPIXMarker( unsigned long color, const char *szName ) = 0;
+	virtual void SetPIXMarker( uint32_t color, const char *szName ) = 0;
 
 	// Batch API
 	// from changelist 166623:
@@ -1433,8 +1520,221 @@ public:
 	virtual void SetNonInteractiveTempFullscreenBuffer( ITexture *pTexture, MaterialNonInteractiveMode_t mode ) = 0;
 	virtual void EnableNonInteractiveMode( MaterialNonInteractiveMode_t mode ) = 0;
 	virtual void RefreshFrontBufferNonInteractive() = 0;
+	// Allocates temp render data. Renderdata goes out of scope at frame end in multicore
+	// Renderdata goes out of scope after refcount goes to zero in singlecore.
+	// Locking/unlocking increases + decreases refcount
+	virtual void *			LockRenderData( int nSizeInBytes ) = 0;
+	virtual void			UnlockRenderData( void *pData ) = 0;
+
+	// Typed version. If specified, pSrcData is copied into the locked memory.
+	template< class E > E*  LockRenderDataTyped( int nCount, const E* pSrcData = NULL );
+
+	// Temp render data gets immediately freed after it's all unlocked in single core.
+	// This prevents it from being freed
+	virtual void			AddRefRenderData() = 0;	
+	virtual void			ReleaseRenderData() = 0;
+
+	// Returns whether a pointer is render data. NOTE: passing NULL returns true
+	virtual bool			IsRenderData( const void *pData ) const = 0;
+	virtual void			PrintfVA( char *fmt, va_list vargs ) = 0;
+	virtual void			Printf( PRINTF_FORMAT_STRING const char *fmt, ... ) = 0;
+	virtual float			Knob( char *knobname, float *setvalue = NULL ) = 0;
+	// Allows us to override the alpha write setting of a material
+	virtual void OverrideAlphaWriteEnable( bool bEnable, bool bAlphaWriteEnable ) = 0;
+	virtual void OverrideColorWriteEnable( bool bOverrideEnable, bool bColorWriteEnable ) = 0;
+
+	virtual void ClearBuffersObeyStencilEx( bool bClearColor, bool bClearAlpha, bool bClearDepth ) = 0;
+
+	// Create a texture from the specified src render target, then call pRecipient->OnAsyncCreateComplete from the main thread.
+	// The texture will be created using the destination format, and will optionally have mipmaps generated.
+	// In case of error, the provided callback function will be called with the error texture.
+	virtual void AsyncCreateTextureFromRenderTarget( ITexture* pSrcRt, const char* pDstName, ImageFormat dstFmt, bool bGenMips, int nAdditionalCreationFlags, IAsyncTextureOperationReceiver* pRecipient, void* pExtraArgs ) = 0;
 };
 
+template< class E > inline E* IMatRenderContext::LockRenderDataTyped( int nCount, const E* pSrcData )
+{
+	int nSizeInBytes = nCount * sizeof(E);
+	E *pDstData = (E*)LockRenderData( nSizeInBytes );
+	if ( pSrcData && pDstData )
+	{
+		memcpy( pDstData, pSrcData, nSizeInBytes );
+	}
+	return pDstData;
+}
+
+
+//-----------------------------------------------------------------------------
+// Utility class for addreffing/releasing render data (prevents freeing on single core)
+//-----------------------------------------------------------------------------
+class CMatRenderDataReference
+{
+public:
+	CMatRenderDataReference();
+	CMatRenderDataReference( IMatRenderContext* pRenderContext );
+	~CMatRenderDataReference();
+	void Lock( IMatRenderContext *pRenderContext );
+	void Release();
+
+private:
+	IMatRenderContext *m_pRenderContext;
+};
+
+
+inline CMatRenderDataReference::CMatRenderDataReference()
+{
+	m_pRenderContext = NULL;
+}
+
+inline CMatRenderDataReference::CMatRenderDataReference( IMatRenderContext* pRenderContext )
+{
+	m_pRenderContext = NULL;
+	Lock( pRenderContext );
+}
+
+inline CMatRenderDataReference::~CMatRenderDataReference()
+{
+	Release();
+}
+
+inline void CMatRenderDataReference::Lock( IMatRenderContext* pRenderContext )
+{
+	if ( !m_pRenderContext )
+	{
+		m_pRenderContext = pRenderContext;
+		m_pRenderContext->AddRefRenderData( );
+	}
+}
+
+inline void CMatRenderDataReference::Release()
+{
+	if ( m_pRenderContext )
+	{
+		m_pRenderContext->ReleaseRenderData( );
+		m_pRenderContext = NULL;
+	}
+}
+
+
+//-----------------------------------------------------------------------------
+// Utility class for locking/unlocking render data
+//-----------------------------------------------------------------------------
+template< typename E > 
+class CMatRenderData
+{
+public:
+	CMatRenderData( IMatRenderContext* pRenderContext );
+	CMatRenderData( IMatRenderContext* pRenderContext, int nCount, const E *pSrcData = NULL );
+	~CMatRenderData();
+	E* Lock( int nCount, const E* pSrcData = NULL ); 
+	void Release();
+	bool IsValid() const;
+	const E* Base() const;
+	E* Base();
+	const E& operator[]( int i ) const;
+	E& operator[]( int i );
+
+private:
+	IMatRenderContext* m_pRenderContext;
+	E *m_pRenderData;
+	int m_nCount;
+	bool m_bNeedsUnlock;
+};
+
+template< typename E >
+inline CMatRenderData<E>::CMatRenderData( IMatRenderContext* pRenderContext )
+{
+	m_pRenderContext = pRenderContext;
+	m_nCount = 0;
+	m_pRenderData = 0;
+	m_bNeedsUnlock = false;
+}
+
+template< typename E >
+inline CMatRenderData<E>::CMatRenderData( IMatRenderContext* pRenderContext, int nCount, const E* pSrcData )
+{
+	m_pRenderContext = pRenderContext;
+	m_nCount = 0;
+	m_pRenderData = 0;
+	m_bNeedsUnlock = false;
+	Lock( nCount, pSrcData );
+}
+
+template< typename E >
+inline CMatRenderData<E>::~CMatRenderData()
+{
+	Release();
+}
+
+template< typename E >
+inline bool CMatRenderData<E>::IsValid() const
+{
+	return m_pRenderData != NULL;
+}
+
+template< typename E >
+inline E* CMatRenderData<E>::Lock( int nCount, const E* pSrcData )
+{
+	m_nCount = nCount;
+	if ( pSrcData && m_pRenderContext->IsRenderData( pSrcData ) )
+	{
+		// Yes, we're const-casting away, but that should be ok since 
+		// the src data is render data
+		m_pRenderData = const_cast<E*>( pSrcData );
+		m_pRenderContext->AddRefRenderData();
+		m_bNeedsUnlock = false;
+		return m_pRenderData;
+	}
+	m_pRenderData = m_pRenderContext->LockRenderDataTyped<E>( nCount, pSrcData );
+	m_bNeedsUnlock = true;
+	return m_pRenderData;
+}
+
+template< typename E >
+inline void CMatRenderData<E>::Release()
+{
+	if ( m_pRenderContext && m_pRenderData )
+	{
+		if ( m_bNeedsUnlock )
+		{
+			m_pRenderContext->UnlockRenderData( m_pRenderData );
+		}
+		else
+		{
+			m_pRenderContext->ReleaseRenderData();
+		}
+	}
+	m_pRenderData = NULL;
+	m_nCount = 0;
+	m_bNeedsUnlock = false;
+}
+
+template< typename E >
+inline E* CMatRenderData<E>::Base()
+{
+	return m_pRenderData;
+}
+
+template< typename E >
+inline const E* CMatRenderData<E>::Base() const
+{
+	return m_pRenderData;
+}
+
+template< typename E >
+inline E& CMatRenderData<E>::operator[]( int i )
+{
+	Assert( ( i >= 0 ) && ( i < m_nCount ) );
+	return m_pRenderData[i];
+}
+
+template< typename E >
+inline const E& CMatRenderData<E>::operator[]( int i ) const
+{
+	Assert( ( i >= 0 ) && ( i < m_nCount ) );
+	return m_pRenderData[i];
+}
+
+
 //-----------------------------------------------------------------------------
 
 class CMatRenderContextPtr : public CRefPtr<IMatRenderContext>
@@ -1468,7 +1768,7 @@ private:
 class PIXEvent
 {
 public:
-	PIXEvent( IMatRenderContext *pRenderContext, const char *szName, unsigned long color = PIX_VALVE_ORANGE )
+	PIXEvent( IMatRenderContext *pRenderContext, const char *szName, uint32_t color = PIX_VALVE_ORANGE )
 	{
 		m_pRenderContext = pRenderContext;
 		Assert( m_pRenderContext );
@@ -1484,7 +1784,8 @@ private:
 };
 
 
-#define PIX_ENABLE 0		// set this to 1 and build engine/studiorender to enable pix events in the engine
+// Also be sure to enable PIX_INSTRUMENTATION in shaderdevicedx8.h
+//#define PIX_ENABLE 1		// set this to 1 and build engine/studiorender to enable pix events in the engine
 
 #if PIX_ENABLE
 #	define PIXEVENT PIXEvent _pixEvent

public/materialsystem/imaterialvar.h

@@ -21,8 +21,8 @@ class VMatrix;
 class ITexture;
 
 #define MAKE_MATERIALVAR_FOURCC(ch0, ch1, ch2, ch3)             \
-		((unsigned long)(ch0) | ((unsigned long)(ch1) << 8) |   \
-		((unsigned long)(ch2) << 16) | ((unsigned long)(ch3) << 24 ))
+		((uint32_t)(ch0) | ((uint32_t)(ch1) << 8) |   \
+		((uint32_t)(ch2) << 16) | ((uint32_t)(ch3) << 24 ))
 
 // This fourcc is reserved.
 #define FOURCC_UNKNOWN	MAKE_MATERIALVAR_FOURCC('U','N','K','N')
@@ -49,7 +49,7 @@ typedef unsigned short MaterialVarSym_t;
 class IMaterialVar
 {
 public:
-	typedef unsigned long FourCC;
+	typedef uint32_t FourCC;
 	
 protected:
 	// base data and accessors

public/mathlib/compressed_3d_unitvec.h

@@ -8,6 +8,7 @@
 #ifndef _3D_UNITVEC_H
 #define _3D_UNITVEC_H
 
+#include <cstdint>
 
 #define UNITVEC_DECLARE_STATICS \
    float cUnitVector::mUVAdjustment[0x2000]; \
@@ -75,8 +76,8 @@ public:
       // a little slower... old pack was 4 multiplies and 2 adds.
       // This is 2 multiplies, 2 adds, and a divide....
       float w = 126.0f / ( tmp.x + tmp.y + tmp.z );
-      long xbits = (long)( tmp.x * w );
-      long ybits = (long)( tmp.y * w );
+      int32_t xbits = (int32_t)( tmp.x * w );
+      int32_t ybits = (int32_t)( tmp.y * w );
 
       Assert( xbits <  127 );
       Assert( xbits >= 0   );
@@ -110,8 +111,8 @@ public:
       // multiplication
 
       // get the x and y bits
-      long xbits = (( mVec & TOP_MASK ) >> 7 );
-      long ybits = ( mVec & BOTTOM_MASK );
+      int32_t xbits = (( mVec & TOP_MASK ) >> 7 );
+      int32_t ybits = ( mVec & BOTTOM_MASK );
 
       // map the numbers back to the triangle (0,0)-(0,126)-(126,0)
       if (( xbits + ybits ) >= 127 )
@@ -139,8 +140,8 @@ public:
    {
       for ( int idx = 0; idx < 0x2000; idx++ )
       {
-         long xbits = idx >> 7;
-         long ybits = idx & BOTTOM_MASK;
+         int32_t xbits = idx >> 7;
+         int32_t ybits = idx & BOTTOM_MASK;
 
          // map the numbers back to the triangle (0,0)-(0,127)-(127,0)
          if (( xbits + ybits ) >= 127 )

public/mathlib/compressed_vector.h

@@ -58,8 +58,8 @@ inline Vector32& Vector32::operator=(const Vector &vOther)
 
 	static float expScale[4] = { 4.0f, 16.0f, 32.f, 64.f };
 
-	float fmax = MAX( fabs( vOther.x ), fabs( vOther.y ) );
-	fmax = fpmax( fmax, fabs( vOther.z ) );
+	float fmax = Max( fabs( vOther.x ), fabs( vOther.y ) );
+	fmax = Max( fmax, (float)fabs( vOther.z ) );
 
 	for (exp = 0; exp < 3; exp++)
 	{
@@ -70,9 +70,9 @@ inline Vector32& Vector32::operator=(const Vector &vOther)
 
 	float fexp = 512.0f / expScale[exp];
 
-	x = clamp( (int)(vOther.x * fexp) + 512, 0, 1023 );
-	y = clamp( (int)(vOther.y * fexp) + 512, 0, 1023 );
-	z = clamp( (int)(vOther.z * fexp) + 512, 0, 1023 );
+	x = Clamp( (int)(vOther.x * fexp) + 512, 0, 1023 );
+	y = Clamp( (int)(vOther.y * fexp) + 512, 0, 1023 );
+	z = Clamp( (int)(vOther.z * fexp) + 512, 0, 1023 );
 	return *this; 
 }
 
@@ -118,8 +118,8 @@ inline Normal32& Normal32::operator=(const Vector &vOther)
 {
 	CHECK_VALID(vOther);
 
-	x = clamp( (int)(vOther.x * 16384) + 16384, 0, 32767 );
-	y = clamp( (int)(vOther.y * 16384) + 16384, 0, 32767 );
+	x = Clamp( (int)(vOther.x * 16384) + 16384, 0, 32767 );
+	y = Clamp( (int)(vOther.y * 16384) + 16384, 0, 32767 );
 	zneg = (vOther.z < 0);
 	//x = vOther.x; 
 	//y = vOther.y; 
@@ -182,9 +182,9 @@ inline Quaternion64& Quaternion64::operator=(const Quaternion &vOther)
 {
 	CHECK_VALID(vOther);
 
-	x = clamp( (int)(vOther.x * 1048576) + 1048576, 0, 2097151 );
-	y = clamp( (int)(vOther.y * 1048576) + 1048576, 0, 2097151 );
-	z = clamp( (int)(vOther.z * 1048576) + 1048576, 0, 2097151 );
+	x = Clamp( (int)(vOther.x * 1048576) + 1048576, 0, 2097151 );
+	y = Clamp( (int)(vOther.y * 1048576) + 1048576, 0, 2097151 );
+	z = Clamp( (int)(vOther.z * 1048576) + 1048576, 0, 2097151 );
 	wneg = (vOther.w < 0);
 	return *this; 
 }
@@ -229,9 +229,9 @@ inline Quaternion48& Quaternion48::operator=(const Quaternion &vOther)
 {
 	CHECK_VALID(vOther);
 
-	x = clamp( (int)(vOther.x * 32768) + 32768, 0, 65535 );
-	y = clamp( (int)(vOther.y * 32768) + 32768, 0, 65535 );
-	z = clamp( (int)(vOther.z * 16384) + 16384, 0, 32767 );
+	x = Clamp( (int)(vOther.x * 32768) + 32768, 0, 65535 );
+	y = Clamp( (int)(vOther.y * 32768) + 32768, 0, 65535 );
+	z = Clamp( (int)(vOther.z * 16384) + 16384, 0, 32767 );
 	wneg = (vOther.w < 0);
 	return *this; 
 }
@@ -276,9 +276,9 @@ inline Quaternion32& Quaternion32::operator=(const Quaternion &vOther)
 {
 	CHECK_VALID(vOther);
 
-	x = clamp( (int)(vOther.x * 1024) + 1024, 0, 2047 );
-	y = clamp( (int)(vOther.y * 512) + 512, 0, 1023 );
-	z = clamp( (int)(vOther.z * 512) + 512, 0, 1023 );
+	x = Clamp( (int)(vOther.x * 1024) + 1024, 0, 2047 );
+	y = Clamp( (int)(vOther.y * 512) + 512, 0, 1023 );
+	z = Clamp( (int)(vOther.z * 512) + 512, 0, 1023 );
 	wneg = (vOther.w < 0);
 	return *this; 
 }

public/mathlib/lightdesc.h

@@ -28,6 +28,7 @@ enum LightType_OptimizationFlags_t
 	LIGHTTYPE_OPTIMIZATIONFLAGS_HAS_ATTENUATION0 = 1,
 	LIGHTTYPE_OPTIMIZATIONFLAGS_HAS_ATTENUATION1 = 2,
 	LIGHTTYPE_OPTIMIZATIONFLAGS_HAS_ATTENUATION2 = 4,
+	LIGHTTYPE_OPTIMIZATIONFLAGS_DERIVED_VALUES_CALCED = 8,
 };
 
 struct LightDesc_t 
@@ -102,6 +103,11 @@ public:
 	{
 		return ((m_Type!=MATERIAL_LIGHT_SPOT) || (rdir.Dot(m_Direction)>=m_PhiDot));
 	}
+
+	float OneOverThetaDotMinusPhiDot() const
+	{
+		return OneOver_ThetaDot_Minus_PhiDot;
+	}
 };
 
 

public/mathlib/math_pfns.h

@@ -28,6 +28,14 @@ extern float (*pfFastCos)(float x);
 #define FastSinCos(x,s,c)   (*pfFastSinCos)(x,s,c)
 #define FastCos(x)			(*pfFastCos)(x)
 
+#if defined(__i386__) || defined(_M_IX86)
+// On x86, the inline FPU or SSE sqrt instruction is faster than
+// the overhead of setting up a function call and saving/restoring
+// the FPU or SSE register state and can be scheduled better, too.
+#undef FastSqrt
+#define FastSqrt(x)			::sqrtf(x)
+#endif
+
 #endif // !_X360
 
 #if defined( _X360 )

public/mathlib/mathlib.h

@@ -1,4 +1,4 @@
-//===== Copyright © 1996-2005, Valve Corporation, All rights reserved. ======//
+//===== Copyright <EFBFBD> 1996-2005, Valve Corporation, All rights reserved. ======//
 //
 // Purpose: 
 //
@@ -7,6 +7,7 @@
 #ifndef MATH_LIB_H
 #define MATH_LIB_H
 
+#include <cmath>
 #include <math.h>
 #include "tier0/basetypes.h"
 #include "mathlib/vector.h"
@@ -15,6 +16,61 @@
 
 #include "mathlib/math_pfns.h"
 
+#if defined(__i386__) || defined(_M_IX86)
+// For MMX intrinsics
+#include <xmmintrin.h>
+#endif
+
+// XXX remove me
+#undef clamp
+
+#ifdef DEBUG  // stop crashing edit-and-continue
+FORCEINLINE float clamp( float val, float minVal, float maxVal )
+{
+	if ( maxVal < minVal )
+		return maxVal;
+	else if( val < minVal )
+		return minVal;
+	else if( val > maxVal )
+		return maxVal;
+	else
+		return val;
+}
+#else // DEBUG
+FORCEINLINE float clamp( float val, float minVal, float maxVal )
+{
+#if defined(__i386__) || defined(_M_IX86)
+	_mm_store_ss( &val,
+		_mm_min_ss(
+			_mm_max_ss(
+				_mm_load_ss(&val),
+				_mm_load_ss(&minVal) ),
+			_mm_load_ss(&maxVal) ) );
+#else
+	val = fpmax(minVal, val);
+	val = fpmin(maxVal, val);
+#endif
+	return val;
+}
+#endif // DEBUG
+
+//
+// Returns a clamped value in the range [min, max].
+//
+template< class T >
+inline T clamp( T const &val, T const &minVal, T const &maxVal )
+{
+	if ( maxVal < minVal )
+		return maxVal;
+	else if( val < minVal )
+		return minVal;
+	else if( val > maxVal )
+		return maxVal;
+	else
+		return val;
+}
+
+
 // plane_t structure
 // !!! if this is changed, it must be changed in asm code too !!!
 // FIXME: does the asm code even exist anymore?
@@ -224,12 +280,12 @@ FORCEINLINE void VectorClear(vec_t *a)
 
 FORCEINLINE float VectorMaximum(const vec_t *v)
 {
-	return MAX( v[0], MAX( v[1], v[2] ) );
+	return V_max( v[0], V_max( v[1], v[2] ) );
 }
 
 FORCEINLINE float VectorMaximum(const Vector& v)
 {
-	return MAX( v.x, MAX( v.y, v.z ) );
+	return V_max( v.x, V_max( v.y, v.z ) );
 }
 
 FORCEINLINE void VectorScale (const float* in, vec_t scale, float* out)
@@ -254,7 +310,7 @@ inline void VectorNegate(vec_t *a)
 }
 
 
-//#define VectorMaximum(a)		( MAX( (a)[0], MAX( (a)[1], (a)[2] ) ) )
+//#define VectorMaximum(a)		( V_max( (a)[0], V_max( (a)[1], (a)[2] ) ) )
 #define Vector2Clear(x)			{(x)[0]=(x)[1]=0;}
 #define Vector2Negate(x)		{(x)[0]=-((x)[0]);(x)[1]=-((x)[1]);}
 #define Vector2Copy(a,b)		{(b)[0]=(a)[0];(b)[1]=(a)[1];}
@@ -281,10 +337,10 @@ FORCEINLINE void VectorMAInline( const Vector& start, float scale, const Vector&
 	dest.z=start.z+direction.z*scale;
 }
 
-//FORCEINLINE void VectorMA( const Vector& start, float scale, const Vector& direction, Vector& dest )
-//{
-//	VectorMAInline(start, scale, direction, dest);
-//}
+FORCEINLINE void VectorMA( const Vector& start, float scale, const Vector& direction, Vector& dest )
+{
+	VectorMAInline(start, scale, direction, dest);
+}
 
 FORCEINLINE void VectorMA( const float * start, float scale, const float *direction, float *dest )
 {
@@ -315,7 +371,7 @@ void inline SinCos( float radians, float *sine, float *cosine )
 {
 #if defined( _X360 )
 	XMScalarSinCos( sine, cosine, radians );
-#elif defined( _WIN32 )
+#elif defined( PLATFORM_WINDOWS_PC32 )
 	_asm
 	{
 		fld		DWORD PTR [radians]
@@ -327,11 +383,12 @@ void inline SinCos( float radians, float *sine, float *cosine )
 		fstp DWORD PTR [edx]
 		fstp DWORD PTR [eax]
 	}
-#elif defined( _LINUX ) || defined ( __APPLE__ )
-	register double __cosr, __sinr;
- 	__asm __volatile__
-    		("fsincos"
-     	: "=t" (__cosr), "=u" (__sinr) : "0" (radians));
+#elif defined( PLATFORM_WINDOWS_PC64 )
+	*sine = sin( radians );
+	*cosine = cos( radians );
+#elif defined( POSIX )
+	double __cosr, __sinr;
+	__asm ("fsincos" : "=t" (__cosr), "=u" (__sinr) : "0" (radians));
 
   	*sine = __sinr;
   	*cosine = __cosr;
@@ -375,11 +432,6 @@ FORCEINLINE T Square( T const &a )
 }
 
 
-FORCEINLINE bool IsPowerOfTwo( uint x )
-{
-	return ( x & ( x - 1 ) ) == 0;
-}
-
 // return the smallest power of two >= x.
 // returns 0 if x == 0 or x > 0x80000000 (ie numbers that would be negative if x was signed)
 // NOTE: the old code took an int, and if you pass in an int of 0x80000000 casted to a uint,
@@ -446,6 +498,19 @@ bool MatricesAreEqual( const matrix3x4_t &src1, const matrix3x4_t &src2, float f
 void MatrixGetColumn( const matrix3x4_t &in, int column, Vector &out );
 void MatrixSetColumn( const Vector &in, int column, matrix3x4_t &out );
 
+inline void MatrixGetTranslation( const matrix3x4_t &in, Vector &out )
+{
+	MatrixGetColumn ( in, 3, out );
+}
+
+inline void MatrixSetTranslation( const Vector &in, matrix3x4_t &out )
+{
+	MatrixSetColumn ( in, 3, out );
+}
+
+void MatrixScaleBy ( const float flScale, matrix3x4_t &out );
+void MatrixScaleByZero ( matrix3x4_t &out );
+
 //void DecomposeRotation( const matrix3x4_t &mat, float *out );
 void ConcatRotations (const matrix3x4_t &in1, const matrix3x4_t &in2, matrix3x4_t &out);
 void ConcatTransforms (const matrix3x4_t &in1, const matrix3x4_t &in2, matrix3x4_t &out);
@@ -606,7 +671,7 @@ template<> FORCEINLINE QAngleByValue Lerp<QAngleByValue>( float flPercent, const
 #endif // VECTOR_NO_SLOW_OPERATIONS
 
 
-// Swap two of anything.
+/// Same as swap(), but won't cause problems with std::swap
 template <class T> 
 FORCEINLINE void V_swap( T& x, T& y )
 {
@@ -621,15 +686,11 @@ template <class T> FORCEINLINE T AVG(T a, T b)
 }
 
 // number of elements in an array of static size
-#define NELEMS(x) ((sizeof(x))/sizeof(x[0]))
+#define NELEMS(x) ARRAYSIZE(x)
 
 // XYZ macro, for printf type functions - ex printf("%f %f %f",XYZ(myvector));
 #define XYZ(v) (v).x,(v).y,(v).z
 
-//
-// Returns a clamped value in the range [min, max].
-//
-#define V_clamp(val, min, max) (((val) > (max)) ? (max) : (((val) < (min)) ? (min) : (val)))
 
 inline float Sign( float x )
 {
@@ -1066,13 +1127,15 @@ inline float SimpleSplineRemapValClamped( float val, float A, float B, float C,
 	if ( A == B )
 		return val >= B ? D : C;
 	float cVal = (val - A) / (B - A);
-	cVal = V_clamp( cVal, 0.0f, 1.0f );
+	cVal = clamp( cVal, 0.0f, 1.0f );
 	return C + (D - C) * SimpleSpline( cVal );
 }
 
 FORCEINLINE int RoundFloatToInt(float f)
 {
-#if defined( _X360 )
+#if defined(__i386__) || defined(_M_IX86) || defined( PLATFORM_WINDOWS_PC64 ) || defined(__x86_64__)
+	return _mm_cvtss_si32(_mm_load_ss(&f));
+#elif defined( _X360 )
 #ifdef Assert
 	Assert( IsFPUControlWordSet() );
 #endif
@@ -1083,63 +1146,18 @@ FORCEINLINE int RoundFloatToInt(float f)
 	};
 	flResult = __fctiw( f );
 	return pResult[1];
-#else // !X360
-	int nResult;
-#if defined( _WIN32 )
-	__asm
-	{
-		fld f
-		fistp nResult
-	}
-#elif defined( _LINUX ) || defined( __APPLE__ )
-	__asm __volatile__ (
-		"fistpl %0;": "=m" (nResult): "t" (f) : "st"
-	);
-#endif
-	return nResult;
+#else
+#error Unknown architecture
 #endif
 }
 
 FORCEINLINE unsigned char RoundFloatToByte(float f)
 {
-#if defined( _X360 )
+	int nResult = RoundFloatToInt(f);
 #ifdef Assert
-	Assert( IsFPUControlWordSet() );
-#endif
-	union
-	{
-		double flResult;
-		int pIntResult[2];
-		unsigned char pResult[8];
-	};
-	flResult = __fctiw( f );
-#ifdef Assert
-	Assert( pIntResult[1] >= 0 && pIntResult[1] <= 255 );
-#endif
-	return pResult[8];
-
-#else // !X360
-	
-	int nResult;
-
-#if defined( _WIN32 )
-	__asm
-	{
-		fld f
-		fistp nResult
-	}
-#elif defined( _LINUX ) || defined( __APPLE__ )
-	__asm __volatile__ (
-		"fistpl %0;": "=m" (nResult): "t" (f) : "st"
-	);
-#endif
-
-#ifdef Assert
-	Assert( nResult >= 0 && nResult <= 255 );
-#endif 
-	return nResult;
-
+	Assert( (nResult & ~0xFF) == 0 );
 #endif
+	return (unsigned char) nResult;
 }
 
 FORCEINLINE unsigned long RoundFloatToUnsignedLong(float f)
@@ -1159,6 +1177,24 @@ FORCEINLINE unsigned long RoundFloatToUnsignedLong(float f)
 	return pResult[1];
 #else  // !X360
 	
+#if defined( PLATFORM_WINDOWS_PC64 )
+	uint nRet = ( uint ) f;
+	if ( nRet & 1 )
+	{
+		if ( ( f - floor( f ) >= 0.5 ) )
+		{
+			nRet++;
+		}
+	}
+	else
+	{
+		if ( ( f - floor( f ) > 0.5 ) )
+		{
+			nRet++;
+		}
+	}
+	return nRet;
+#else // PLATFORM_WINDOWS_PC64
 	unsigned char nResult[8];
 
 	#if defined( _WIN32 )
@@ -1167,14 +1203,15 @@ FORCEINLINE unsigned long RoundFloatToUnsignedLong(float f)
 			fld f
 			fistp       qword ptr nResult
 		}
-#elif defined( _LINUX ) || defined( __APPLE__ )
+	#elif POSIX
 		__asm __volatile__ (
 			"fistpl %0;": "=m" (nResult): "t" (f) : "st"
 		);
 	#endif
 
 		return *((unsigned long*)nResult);
-#endif
+#endif // PLATFORM_WINDOWS_PC64
+#endif // !X360
 }
 
 FORCEINLINE bool IsIntegralValue( float flValue, float flTolerance = 0.001f )
@@ -1194,29 +1231,8 @@ FORCEINLINE int Float2Int( float a )
 	flResult = __fctiwz( a );
 	return pResult[1];
 #else  // !X360
-	
-	int RetVal;
-
-#if defined( _WIN32 )
-	int CtrlwdHolder;
-	int CtrlwdSetter;
-	__asm 
-	{
-		fld    a					// push 'a' onto the FP stack
-		fnstcw CtrlwdHolder		// store FPU control word
-		movzx  eax, CtrlwdHolder	// move and zero extend word into eax
-		and    eax, 0xFFFFF3FF	// set all bits except rounding bits to 1
-		or     eax, 0x00000C00	// set rounding mode bits to round towards zero
-		mov    CtrlwdSetter, eax	// Prepare to set the rounding mode -- prepare to enter plaid!
-		fldcw  CtrlwdSetter		// Entering plaid!
-		fistp  RetVal				// Store and converted (to int) result
-		fldcw  CtrlwdHolder		// Restore control word
-	}
-#elif defined( _LINUX ) || defined ( __APPLE__ ) 
-	RetVal = static_cast<int>( a );
-#endif
-
-	return RetVal;
+	// Rely on compiler to generate CVTTSS2SI on x86
+	return (int) a;
 #endif
 }
 
@@ -1224,46 +1240,45 @@ FORCEINLINE int Float2Int( float a )
 inline int Floor2Int( float a )
 {
 	int RetVal;
-
-#if defined( _X360 )
-	RetVal = (int)floor( a );
-#elif defined( _WIN32 )
-   int CtrlwdHolder;
-   int CtrlwdSetter;
-   __asm 
-   {
-      fld    a					// push 'a' onto the FP stack
-      fnstcw CtrlwdHolder		// store FPU control word
-      movzx  eax, CtrlwdHolder	// move and zero extend word into eax
-      and    eax, 0xFFFFF3FF	// set all bits except rounding bits to 1
-      or     eax, 0x00000400	// set rounding mode bits to round down
-      mov    CtrlwdSetter, eax	// Prepare to set the rounding mode -- prepare to enter plaid!
-      fldcw  CtrlwdSetter		// Entering plaid!
-      fistp  RetVal				// Store floored and converted (to int) result
-      fldcw  CtrlwdHolder		// Restore control word
-   }
-#elif defined( _LINUX ) || defined( __APPLE__ )
+#if defined( __i386__ )
+	// Convert to int and back, compare, subtract one if too big
+	__m128 a128 = _mm_set_ss(a);
+	RetVal = _mm_cvtss_si32(a128);
+    __m128 rounded128 = _mm_cvt_si2ss(_mm_setzero_ps(), RetVal);
+	RetVal -= _mm_comigt_ss( rounded128, a128 );
+#else
 	RetVal = static_cast<int>( floor(a) );
 #endif
-
 	return RetVal;
 }
 
 //-----------------------------------------------------------------------------
 // Fast color conversion from float to unsigned char
 //-----------------------------------------------------------------------------
-FORCEINLINE unsigned char FastFToC( float c )
+FORCEINLINE unsigned int FastFToC( float c )
 {
-	volatile float dc;
-
-	// ieee trick
-	dc = c * 255.0f + (float)(1 << 23);
-	
-	// return the lsb
-#if defined( _X360 )
-	return ((unsigned char*)&dc)[3];
+#if defined( __i386__ )
+	// IEEE float bit manipulation works for values between [0, 1<<23)
+	union { float f; int i; } convert = { c*255.0f + (float)(1<<23) };
+	return convert.i & 255;
 #else
-	return *(unsigned char*)&dc;
+	// consoles CPUs suffer from load-hit-store penalty
+	return Float2Int( c * 255.0f );
+#endif
+}
+
+//-----------------------------------------------------------------------------
+// Fast conversion from float to integer with magnitude less than 2**22
+//-----------------------------------------------------------------------------
+FORCEINLINE int FastFloatToSmallInt( float c )
+{
+#if defined( __i386__ )
+	// IEEE float bit manipulation works for values between [-1<<22, 1<<22)
+	union { float f; int i; } convert = { c + (float)(3<<22) };
+	return (convert.i & ((1<<23)-1)) - (1<<22);
+#else
+	// consoles CPUs suffer from load-hit-store penalty
+	return Float2Int( c );
 #endif
 }
 
@@ -1275,35 +1290,22 @@ FORCEINLINE unsigned char FastFToC( float c )
 inline float ClampToMsec( float in )
 {
 	int msec = Floor2Int( in * 1000.0f + 0.5f );
-	return msec / 1000.0f;
+	return 0.001f * msec;
 }
 
 // Over 15x faster than: (int)ceil(value)
 inline int Ceil2Int( float a )
 {
    int RetVal;
-
-#if defined( _X360 )
-	RetVal = (int)ceil( a );
-#elif defined( _WIN32 )
-   int CtrlwdHolder;
-   int CtrlwdSetter;
-   __asm 
-   {
-      fld    a					// push 'a' onto the FP stack
-      fnstcw CtrlwdHolder		// store FPU control word
-      movzx  eax, CtrlwdHolder	// move and zero extend word into eax
-      and    eax, 0xFFFFF3FF	// set all bits except rounding bits to 1
-      or     eax, 0x00000800	// set rounding mode bits to round down
-      mov    CtrlwdSetter, eax	// Prepare to set the rounding mode -- prepare to enter plaid!
-      fldcw  CtrlwdSetter		// Entering plaid!
-      fistp  RetVal				// Store floored and converted (to int) result
-      fldcw  CtrlwdHolder		// Restore control word
-   }
-#elif defined( _LINUX ) || defined( __APPLE__ )
+#if defined( __i386__ )
+   // Convert to int and back, compare, add one if too small
+   __m128 a128 = _mm_load_ss(&a);
+   RetVal = _mm_cvtss_si32(a128);
+   __m128 rounded128 = _mm_cvt_si2ss(_mm_setzero_ps(), RetVal);
+   RetVal += _mm_comilt_ss( rounded128, a128 );
+#else
    RetVal = static_cast<int>( ceil(a) );
 #endif
-
 	return RetVal;
 }
 
@@ -1419,7 +1421,7 @@ FORCEINLINE unsigned char LinearToLightmap( float f )
 
 FORCEINLINE void ColorClamp( Vector& color )
 {
-	float maxc = MAX( color.x, MAX( color.y, color.z ) );
+	float maxc = V_max( color.x, V_max( color.y, color.z ) );
 	if ( maxc > 1.0f )
 	{
 		float ooMax = 1.0f / maxc;
@@ -1915,10 +1917,10 @@ FORCEINLINE unsigned int * PackNormal_SHORT2( float nx, float ny, float nz, unsi
 	ny *= 16384.0f;
 
 	// '0' and '32768' values are invalid encodings
-	nx = MAX( nx, 1.0f );		// Make sure there are no zero values
-	ny = MAX( ny, 1.0f );
-	nx = MIN( nx, 32767.0f );	// Make sure there are no 32768 values
-	ny = MIN( ny, 32767.0f );
+	nx = V_max( nx, 1.0f );		// Make sure there are no zero values
+	ny = V_max( ny, 1.0f );
+	nx = V_min( nx, 32767.0f );	// Make sure there are no 32768 values
+	ny = V_min( ny, 32767.0f );
 
 	if ( nz < 0.0f )
 		nx = -nx;				// Set the sign bit for z
@@ -2068,6 +2070,46 @@ void RGBtoHSV( const Vector &rgb, Vector &hsv );
 void HSVtoRGB( const Vector &hsv, Vector &rgb );
 
 
+//-----------------------------------------------------------------------------
+// Fast version of pow and log
+//-----------------------------------------------------------------------------
+
+float FastLog2(float i);			// log2( i )
+float FastPow2(float i);			// 2^i
+float FastPow(float a, float b);	// a^b
+float FastPow10( float i );			// 10^i
+
+//-----------------------------------------------------------------------------
+// For testing float equality
+//-----------------------------------------------------------------------------
+
+inline bool CloseEnough( float a, float b, float epsilon = EQUAL_EPSILON )
+{
+	return fabs( a - b ) <= epsilon;
+}
+
+inline bool CloseEnough( const Vector &a, const Vector &b, float epsilon = EQUAL_EPSILON )
+{
+	return fabs( a.x - b.x ) <= epsilon &&
+		fabs( a.y - b.y ) <= epsilon &&
+		fabs( a.z - b.z ) <= epsilon;
+}
+
+// Fast compare
+// maxUlps is the maximum error in terms of Units in the Last Place. This 
+// specifies how big an error we are willing to accept in terms of the value
+// of the least significant digit of the floating point number’s 
+// representation. maxUlps can also be interpreted in terms of how many 
+// representable floats we are willing to accept between A and B. 
+// This function will allow maxUlps-1 floats between A and B.
+bool AlmostEqual(float a, float b, int maxUlps = 10);
+
+inline bool AlmostEqual( const Vector &a, const Vector &b, int maxUlps = 10)
+{
+	return AlmostEqual( a.x, b.x, maxUlps ) &&
+		AlmostEqual( a.y, b.y, maxUlps ) &&
+		AlmostEqual( a.z, b.z, maxUlps );
+}
 
 #endif	// MATH_BASE_H
 

public/mathlib/matrixmath.h

@@ -0,0 +1,385 @@
+//========= Copyright Valve Corporation, All rights reserved. ============//
+//
+// Purpose: 
+//
+//  A set of generic, template-based matrix functions.
+//===========================================================================//
+
+#ifndef MATRIXMATH_H
+#define MATRIXMATH_H
+
+#include <stdarg.h>
+
+// The operations in this file can perform basic matrix operations on matrices represented
+// using any class that supports the necessary operations:
+//
+//  .Element( row, col )  - return the element at a given matrox position
+//  .SetElement( row, col, val ) - modify an element
+//  .Width(), .Height() - get dimensions
+//  .SetDimensions( nrows, ncols) - set a matrix to be un-initted and the appropriate size
+//
+// Generally, vectors can be used with these functions by using N x 1 matrices to represent them.
+//  Matrices are addressed as row, column, and indices are 0-based
+//
+//
+// Note that the template versions of these routines are defined for generality - it is expected
+// that template specialization is used for common high performance cases.
+
+namespace MatrixMath
+{
+	/// M *= flScaleValue
+	template<class MATRIXCLASS>
+	void ScaleMatrix( MATRIXCLASS &matrix, float flScaleValue )
+	{
+		for( int i = 0; i < matrix.Height(); i++ )
+		{
+			for( int j = 0; j < matrix.Width(); j++ )
+			{
+				matrix.SetElement( i, j, flScaleValue * matrix.Element( i, j ) );
+			}
+		}
+	}
+
+	/// AppendElementToMatrix - same as setting the element, except only works when all calls
+	/// happen in top to bottom left to right order, end you have to call FinishedAppending when
+	/// done. For normal matrix classes this is not different then SetElement, but for
+	/// CSparseMatrix, it is an accelerated way to fill a matrix from scratch.
+	template<class MATRIXCLASS>
+	FORCEINLINE void AppendElement( MATRIXCLASS &matrix, int nRow, int nCol, float flValue )
+	{
+		matrix.SetElement( nRow, nCol, flValue );			// default implementation
+	}
+
+	template<class MATRIXCLASS>
+	FORCEINLINE void FinishedAppending( MATRIXCLASS &matrix ) {} // default implementation
+
+	/// M += fl
+	template<class MATRIXCLASS>
+	void AddToMatrix( MATRIXCLASS &matrix, float flAddend )
+	{
+		for( int i = 0; i < matrix.Height(); i++ )
+		{
+			for( int j = 0; j < matrix.Width(); j++ )
+			{
+				matrix.SetElement( i, j, flAddend + matrix.Element( i, j ) );
+			}
+		}
+	}
+
+	/// transpose
+	template<class MATRIXCLASSIN, class MATRIXCLASSOUT>
+	void TransposeMatrix( MATRIXCLASSIN const &matrixIn, MATRIXCLASSOUT *pMatrixOut )
+	{
+		pMatrixOut->SetDimensions( matrixIn.Width(), matrixIn.Height() );
+		for( int i = 0; i < pMatrixOut->Height(); i++ )
+		{
+			for( int j = 0; j < pMatrixOut->Width(); j++ )
+			{
+				AppendElement( *pMatrixOut, i, j, matrixIn.Element( j, i ) );
+			}
+		}
+		FinishedAppending( *pMatrixOut );
+	}
+
+	/// copy
+	template<class MATRIXCLASSIN, class MATRIXCLASSOUT>
+	void CopyMatrix( MATRIXCLASSIN const &matrixIn, MATRIXCLASSOUT *pMatrixOut )
+	{
+		pMatrixOut->SetDimensions( matrixIn.Height(), matrixIn.Width() );
+		for( int i = 0; i < matrixIn.Height(); i++ )
+		{
+			for( int j = 0; j < matrixIn.Width(); j++ )
+			{
+				AppendElement( *pMatrixOut, i, j, matrixIn.Element( i, j ) );
+			}
+		}
+		FinishedAppending( *pMatrixOut );
+	}
+
+
+
+	/// M+=M
+	template<class MATRIXCLASSIN, class MATRIXCLASSOUT>
+	void AddMatrixToMatrix( MATRIXCLASSIN const &matrixIn, MATRIXCLASSOUT *pMatrixOut )
+	{
+		for( int i = 0; i < matrixIn.Height(); i++ )
+		{
+			for( int j = 0; j < matrixIn.Width(); j++ )
+			{
+				pMatrixOut->SetElement( i, j, pMatrixOut->Element( i, j ) + matrixIn.Element( i, j ) );
+			}
+		}
+	}
+
+	// M += scale * M
+	template<class MATRIXCLASSIN, class MATRIXCLASSOUT>
+	void AddScaledMatrixToMatrix( float flScale, MATRIXCLASSIN const &matrixIn, MATRIXCLASSOUT *pMatrixOut )
+	{
+		for( int i = 0; i < matrixIn.Height(); i++ )
+		{
+			for( int j = 0; j < matrixIn.Width(); j++ )
+			{
+				pMatrixOut->SetElement( i, j, pMatrixOut->Element( i, j ) + flScale * matrixIn.Element( i, j ) );
+			}
+		}
+	}
+
+
+	// simple way to initialize a matrix with constants from code.
+	template<class MATRIXCLASSOUT> 
+	void SetMatrixToIdentity( MATRIXCLASSOUT *pMatrixOut, float flDiagonalValue = 1.0 )
+	{
+		for( int i = 0; i < pMatrixOut->Height(); i++ )
+		{
+			for( int j = 0; j < pMatrixOut->Width(); j++ )
+			{
+				AppendElement( *pMatrixOut, i, j, ( i == j ) ? flDiagonalValue : 0 );
+			}
+		}
+		FinishedAppending( *pMatrixOut );
+	}
+
+	//// simple way to initialize a matrix with constants from code
+	template<class MATRIXCLASSOUT> 
+	void SetMatrixValues( MATRIXCLASSOUT *pMatrix, int nRows, int nCols, ... )
+	{
+		va_list argPtr;
+		va_start( argPtr, nCols );
+
+		pMatrix->SetDimensions( nRows, nCols );
+		for( int nRow = 0; nRow < nRows; nRow++ )
+		{
+			for( int nCol = 0; nCol < nCols; nCol++ )
+			{
+				double flNewValue = va_arg( argPtr, double );
+				pMatrix->SetElement( nRow, nCol, flNewValue );
+			}
+		}
+		va_end( argPtr );
+	}
+
+
+	/// row and colum accessors. treat a row or a column as a column vector
+	template<class MATRIXTYPE> class MatrixRowAccessor
+	{
+	public:
+		FORCEINLINE MatrixRowAccessor( MATRIXTYPE const &matrix, int nRow )
+		{
+			m_pMatrix = &matrix;
+			m_nRow = nRow;
+		}
+
+		FORCEINLINE float Element( int nRow, int nCol ) const
+		{
+			Assert( nCol == 0 );
+			return m_pMatrix->Element( m_nRow, nRow );
+		}
+
+		FORCEINLINE int Width( void ) const { return 1; };
+		FORCEINLINE int Height( void ) const { return m_pMatrix->Width(); }
+
+	private:
+		MATRIXTYPE const *m_pMatrix;
+		int m_nRow;
+	};
+
+	template<class MATRIXTYPE> class MatrixColumnAccessor
+	{
+	public:
+		FORCEINLINE MatrixColumnAccessor( MATRIXTYPE const &matrix, int nColumn )
+		{
+			m_pMatrix = &matrix;
+			m_nColumn = nColumn;
+		}
+
+		FORCEINLINE float Element( int nRow, int nColumn ) const
+		{
+			Assert( nColumn == 0 );
+			return m_pMatrix->Element( nRow, m_nColumn );
+		}
+
+		FORCEINLINE int Width( void ) const { return 1; }
+		FORCEINLINE int Height( void ) const { return m_pMatrix->Height(); }
+	private:
+		MATRIXTYPE const *m_pMatrix;
+		int m_nColumn;
+	};
+
+	/// this translator acts as a proxy for the transposed matrix
+	template<class MATRIXTYPE> class MatrixTransposeAccessor
+	{
+	public:
+		FORCEINLINE MatrixTransposeAccessor( MATRIXTYPE const & matrix )
+		{
+			m_pMatrix = &matrix;
+		}
+
+		FORCEINLINE float Element( int nRow, int nColumn ) const
+		{
+			return m_pMatrix->Element( nColumn, nRow );
+		}
+
+		FORCEINLINE int Width( void ) const { return m_pMatrix->Height(); }
+		FORCEINLINE int Height( void ) const { return m_pMatrix->Width(); }
+	private:
+		MATRIXTYPE const *m_pMatrix;
+	};
+
+	/// this tranpose returns a wrapper around it's argument, allowing things like AddMatrixToMatrix( Transpose( matA ), &matB ) without an extra copy
+	template<class MATRIXCLASSIN>
+	MatrixTransposeAccessor<MATRIXCLASSIN> TransposeMatrix( MATRIXCLASSIN const &matrixIn )
+	{
+		return MatrixTransposeAccessor<MATRIXCLASSIN>( matrixIn );
+	}
+
+
+	/// retrieve rows and columns
+	template<class MATRIXTYPE>
+	FORCEINLINE MatrixColumnAccessor<MATRIXTYPE> MatrixColumn( MATRIXTYPE const &matrix, int nColumn )
+	{
+		return MatrixColumnAccessor<MATRIXTYPE>( matrix, nColumn );
+	}
+
+	template<class MATRIXTYPE>
+	FORCEINLINE MatrixRowAccessor<MATRIXTYPE> MatrixRow( MATRIXTYPE const &matrix, int nRow )
+	{
+		return MatrixRowAccessor<MATRIXTYPE>( matrix, nRow );
+	}
+
+	//// dot product between vectors (or rows and/or columns via accessors)
+	template<class MATRIXACCESSORATYPE, class MATRIXACCESSORBTYPE >
+	float InnerProduct( MATRIXACCESSORATYPE const &vecA, MATRIXACCESSORBTYPE const &vecB )
+	{
+		Assert( vecA.Width() == 1 );
+		Assert( vecB.Width() == 1 );
+		Assert( vecA.Height() == vecB.Height() );
+		double flResult = 0;
+		for( int i = 0; i < vecA.Height(); i++ )
+		{
+			flResult += vecA.Element( i, 0 ) * vecB.Element( i, 0 );
+		}
+		return flResult;
+	}
+
+
+
+	/// matrix x matrix multiplication
+	template<class MATRIXATYPE, class MATRIXBTYPE, class MATRIXOUTTYPE>
+	void MatrixMultiply( MATRIXATYPE const &matA, MATRIXBTYPE const &matB, MATRIXOUTTYPE *pMatrixOut )
+	{
+		Assert( matA.Width() == matB.Height() );
+		pMatrixOut->SetDimensions( matA.Height(), matB.Width() );
+		for( int i = 0; i < matA.Height(); i++ )
+		{
+			for( int j = 0; j < matB.Width(); j++ )
+			{
+				pMatrixOut->SetElement( i, j, InnerProduct( MatrixRow( matA, i ), MatrixColumn( matB, j ) ) );
+			}
+		}
+	}
+
+	/// solve Ax=B via the conjugate graident method. Code and naming conventions based on the
+	/// wikipedia article.
+	template<class ATYPE, class XTYPE, class BTYPE>
+	void ConjugateGradient( ATYPE const &matA, BTYPE const &vecB, XTYPE &vecX, float flTolerance = 1.0e-20 )
+	{
+		XTYPE vecR;
+		vecR.SetDimensions( vecX.Height(), 1 );
+		MatrixMultiply( matA, vecX, &vecR );
+		ScaleMatrix( vecR, -1 );
+		AddMatrixToMatrix( vecB, &vecR );
+		XTYPE vecP;
+		CopyMatrix( vecR, &vecP );
+		float flRsOld = InnerProduct( vecR, vecR );
+		for( int nIter = 0; nIter < 100; nIter++ )
+		{
+			XTYPE vecAp;
+			MatrixMultiply( matA, vecP, &vecAp );
+			float flDivisor = InnerProduct( vecAp, vecP );
+			float flAlpha = flRsOld / flDivisor;
+			AddScaledMatrixToMatrix( flAlpha, vecP, &vecX );
+			AddScaledMatrixToMatrix( -flAlpha, vecAp, &vecR );
+			float flRsNew = InnerProduct( vecR, vecR );
+			if ( flRsNew < flTolerance )
+			{
+				break;
+			}
+			ScaleMatrix( vecP, flRsNew / flRsOld );
+			AddMatrixToMatrix( vecR, &vecP );
+			flRsOld = flRsNew;
+		}
+	}
+
+	/// solve (A'*A) x=B via the conjugate gradient method. Code and naming conventions based on
+	/// the wikipedia article. Same as Conjugate gradient but allows passing in two matrices whose
+	/// product is used as the A matrix (in order to preserve sparsity)
+	template<class ATYPE, class APRIMETYPE, class XTYPE, class BTYPE>
+	void ConjugateGradient( ATYPE const &matA, APRIMETYPE const &matAPrime, BTYPE const &vecB, XTYPE &vecX, float flTolerance = 1.0e-20 )
+	{
+		XTYPE vecR1;
+		vecR1.SetDimensions( vecX.Height(), 1 );
+		MatrixMultiply( matA, vecX, &vecR1 );
+		XTYPE vecR;
+		vecR.SetDimensions( vecR1.Height(), 1 );
+		MatrixMultiply( matAPrime, vecR1, &vecR );
+		ScaleMatrix( vecR, -1 );
+		AddMatrixToMatrix( vecB, &vecR );
+		XTYPE vecP;
+		CopyMatrix( vecR, &vecP );
+		float flRsOld = InnerProduct( vecR, vecR );
+		for( int nIter = 0; nIter < 100; nIter++ )
+		{
+			XTYPE vecAp1;
+			MatrixMultiply( matA, vecP, &vecAp1 );
+			XTYPE vecAp;
+			MatrixMultiply( matAPrime, vecAp1, &vecAp );
+			float flDivisor = InnerProduct( vecAp, vecP );
+			float flAlpha = flRsOld / flDivisor;
+			AddScaledMatrixToMatrix( flAlpha, vecP, &vecX );
+			AddScaledMatrixToMatrix( -flAlpha, vecAp, &vecR );
+			float flRsNew = InnerProduct( vecR, vecR );
+			if ( flRsNew < flTolerance )
+			{
+				break;
+			}
+			ScaleMatrix( vecP, flRsNew / flRsOld );
+			AddMatrixToMatrix( vecR, &vecP );
+			flRsOld = flRsNew;
+		}
+	}
+
+	
+	template<class ATYPE,  class XTYPE, class BTYPE>
+	void LeastSquaresFit( ATYPE const &matA, BTYPE const &vecB, XTYPE &vecX )
+	{
+		// now, generate the normal equations
+		BTYPE vecBeta;
+		MatrixMath::MatrixMultiply( MatrixMath::TransposeMatrix( matA ), vecB, &vecBeta );
+
+		vecX.SetDimensions( matA.Width(), 1 );
+		MatrixMath::SetMatrixToIdentity( &vecX );
+
+		ATYPE matATransposed;
+		TransposeMatrix( matA, &matATransposed );
+		ConjugateGradient( matA, matATransposed, vecBeta, vecX, 1.0e-20 );
+	}
+
+};
+
+/// a simple fixed-size matrix class
+template<int NUMROWS, int NUMCOLS> class CFixedMatrix
+{
+public:
+	FORCEINLINE int Width( void ) const { return NUMCOLS; }
+	FORCEINLINE int Height( void ) const { return NUMROWS; }
+	FORCEINLINE float Element( int nRow, int nCol ) const { return m_flValues[nRow][nCol]; }
+	FORCEINLINE void SetElement( int nRow, int nCol, float flValue ) { m_flValues[nRow][nCol] = flValue; }
+	FORCEINLINE void SetDimensions( int nNumRows, int nNumCols ) { Assert( ( nNumRows == NUMROWS ) && ( nNumCols == NUMCOLS ) ); }
+
+private:
+	float m_flValues[NUMROWS][NUMCOLS];
+};
+
+
+
+#endif //matrixmath_h

public/mathlib/ssemath.h

@@ -1,4 +1,4 @@
-//===== Copyright © 1996-2005, Valve Corporation, All rights reserved. ======//
+//===== Copyright <EFBFBD> 1996-2005, Valve Corporation, All rights reserved. ======//
 //
 // Purpose: - defines SIMD "structure of arrays" classes and functions.
 //
@@ -15,7 +15,7 @@
 #include <mathlib/vector.h>
 #include <mathlib/mathlib.h>
 
-#if defined(_LINUX) || defined(__APPLE__)
+#if defined(GNUC)
 #define USE_STDC_FOR_SIMD 0
 #else
 #define USE_STDC_FOR_SIMD 0
@@ -108,7 +108,7 @@ struct ALIGN16 intx4
 			m_i32[2] == other.m_i32[2] &&
 			m_i32[3] == other.m_i32[3] 	;
 	}
-};
+} ALIGN16_POST;
 
 
 #if defined( _DEBUG ) && defined( _X360 )
@@ -170,17 +170,17 @@ extern const fltx4 Four_Negative_FLT_MAX;						// -FLT_MAX, -FLT_MAX, -FLT_MAX,
 extern const fltx4 g_SIMD_0123;									// 0 1 2 3 as float
 
 // external aligned integer constants
-extern const ALIGN16 int32 g_SIMD_clear_signmask[];			// 0x7fffffff x 4
-extern const ALIGN16 int32 g_SIMD_signmask[];				// 0x80000000 x 4
-extern const ALIGN16 int32 g_SIMD_lsbmask[];				// 0xfffffffe x 4
-extern const ALIGN16 int32 g_SIMD_clear_wmask[];			// -1 -1 -1 0
-extern const ALIGN16 int32 g_SIMD_ComponentMask[4][4];		// [0xFFFFFFFF 0 0 0], [0 0xFFFFFFFF 0 0], [0 0 0xFFFFFFFF 0], [0 0 0 0xFFFFFFFF]
-extern const ALIGN16 int32 g_SIMD_AllOnesMask[];			// ~0,~0,~0,~0
-extern const ALIGN16 int32 g_SIMD_Low16BitsMask[];			// 0xffff x 4
+extern const ALIGN16 int32 g_SIMD_clear_signmask[] ALIGN16_POST;			// 0x7fffffff x 4
+extern const ALIGN16 int32 g_SIMD_signmask[] ALIGN16_POST;				// 0x80000000 x 4
+extern const ALIGN16 int32 g_SIMD_lsbmask[] ALIGN16_POST;				// 0xfffffffe x 4
+extern const ALIGN16 int32 g_SIMD_clear_wmask[] ALIGN16_POST;			// -1 -1 -1 0
+extern const ALIGN16 int32 g_SIMD_ComponentMask[4][4] ALIGN16_POST;		// [0xFFFFFFFF 0 0 0], [0 0xFFFFFFFF 0 0], [0 0 0xFFFFFFFF 0], [0 0 0 0xFFFFFFFF]
+extern const ALIGN16 int32 g_SIMD_AllOnesMask[] ALIGN16_POST;			// ~0,~0,~0,~0
+extern const ALIGN16 int32 g_SIMD_Low16BitsMask[] ALIGN16_POST;			// 0xffff x 4
 
 // this mask is used for skipping the tail of things. If you have N elements in an array, and wish
 // to mask out the tail, g_SIMD_SkipTailMask[N & 3] what you want to use for the last iteration.
-extern const int32 ALIGN16 g_SIMD_SkipTailMask[4][4];
+extern const int32 ALIGN16 g_SIMD_SkipTailMask[4][4] ALIGN16_POST;
 
 // Define prefetch macros.
 // The characteristics of cache and prefetch are completely 
@@ -436,23 +436,23 @@ FORCEINLINE fltx4 ArcTan2SIMD( const fltx4 &a, const fltx4 &b )
 	return result;
 }
 
-FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b )				// MAX(a,b)
+FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b )				// max(a,b)
 {
 	fltx4 retVal;
-	SubFloat( retVal, 0 ) = MAX( SubFloat( a, 0 ), SubFloat( b, 0 ) );
-	SubFloat( retVal, 1 ) = MAX( SubFloat( a, 1 ), SubFloat( b, 1 ) );
-	SubFloat( retVal, 2 ) = MAX( SubFloat( a, 2 ), SubFloat( b, 2 ) );
-	SubFloat( retVal, 3 ) = MAX( SubFloat( a, 3 ), SubFloat( b, 3 ) );
+	SubFloat( retVal, 0 ) = max( SubFloat( a, 0 ), SubFloat( b, 0 ) );
+	SubFloat( retVal, 1 ) = max( SubFloat( a, 1 ), SubFloat( b, 1 ) );
+	SubFloat( retVal, 2 ) = max( SubFloat( a, 2 ), SubFloat( b, 2 ) );
+	SubFloat( retVal, 3 ) = max( SubFloat( a, 3 ), SubFloat( b, 3 ) );
 	return retVal;
 }
 
-FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b )				// MIN(a,b)
+FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b )				// min(a,b)
 {
 	fltx4 retVal;
-	SubFloat( retVal, 0 ) = MIN( SubFloat( a, 0 ), SubFloat( b, 0 ) );
-	SubFloat( retVal, 1 ) = MIN( SubFloat( a, 1 ), SubFloat( b, 1 ) );
-	SubFloat( retVal, 2 ) = MIN( SubFloat( a, 2 ), SubFloat( b, 2 ) );
-	SubFloat( retVal, 3 ) = MIN( SubFloat( a, 3 ), SubFloat( b, 3 ) );
+	SubFloat( retVal, 0 ) = min( SubFloat( a, 0 ), SubFloat( b, 0 ) );
+	SubFloat( retVal, 1 ) = min( SubFloat( a, 1 ), SubFloat( b, 1 ) );
+	SubFloat( retVal, 2 ) = min( SubFloat( a, 2 ), SubFloat( b, 2 ) );
+	SubFloat( retVal, 3 ) = min( SubFloat( a, 3 ), SubFloat( b, 3 ) );
 	return retVal;
 }
 
@@ -858,7 +858,7 @@ FORCEINLINE void TransposeSIMD( fltx4 & x, fltx4 & y, fltx4 & z, fltx4 & w )
 // and replicate it to the whole return value.
 FORCEINLINE fltx4 FindLowestSIMD3( const fltx4 & a )
 {
-	float lowest = MIN( MIN( SubFloat(a, 0), SubFloat(a, 1) ), SubFloat(a, 2));
+	float lowest = min( min( SubFloat(a, 0), SubFloat(a, 1) ), SubFloat(a, 2));
 	return ReplicateX4(lowest);
 }
 
@@ -866,7 +866,7 @@ FORCEINLINE fltx4 FindLowestSIMD3( const fltx4 & a )
 // and replicate it to the whole return value.
 FORCEINLINE fltx4 FindHighestSIMD3( const fltx4 & a )
 {
-	float highest = MAX( MAX( SubFloat(a, 0), SubFloat(a, 1) ), SubFloat(a, 2));
+	float highest = max( max( SubFloat(a, 0), SubFloat(a, 1) ), SubFloat(a, 2));
 	return ReplicateX4(highest);
 }
 
@@ -1067,12 +1067,12 @@ FORCEINLINE fltx4 ArcTan2SIMD( const fltx4 &a, const fltx4 &b )
 
 // DivSIMD defined further down, since it uses ReciprocalSIMD
 
-FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b )				// MAX(a,b)
+FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b )				// max(a,b)
 {
 	return __vmaxfp( a, b );
 }
 
-FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b )				// MIN(a,b)
+FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b )				// min(a,b)
 {
 	return __vminfp( a, b );
 }
@@ -1520,11 +1520,11 @@ FORCEINLINE fltx4 FindLowestSIMD3( const fltx4 & a )
 	compareOne = __vrlimi( compareOne, a, 8 | 4 , 1 );
 	// compareOne is [y,z,G,G]
 	fltx4 retval = MinSIMD( a, compareOne );
-	// retVal is [MIN(x,y), MIN(y,z), G, G]
+	// retVal is [min(x,y), min(y,z), G, G]
 	compareOne = __vrlimi( compareOne, a, 8 , 2);
 	// compareOne is [z, G, G, G]
 	retval = MinSIMD( retval, compareOne );
-	// retVal = [ MIN(MIN(x,y),z), G, G, G ]
+	// retVal = [ min(min(x,y),z), G, G, G ]
 	
 	// splat the x component out to the whole vector and return
 	return SplatXSIMD( retval );
@@ -1544,11 +1544,11 @@ FORCEINLINE fltx4 FindHighestSIMD3( const fltx4 & a )
 	compareOne = __vrlimi( compareOne, a, 8 | 4 , 1 );
 	// compareOne is [y,z,G,G]
 	fltx4 retval = MaxSIMD( a, compareOne );
-	// retVal is [MAX(x,y), MAX(y,z), G, G]
+	// retVal is [max(x,y), max(y,z), G, G]
 	compareOne = __vrlimi( compareOne, a, 8 , 2);
 	// compareOne is [z, G, G, G]
 	retval = MaxSIMD( retval, compareOne );
-	// retVal = [ MAX(MAX(x,y),z), G, G, G ]
+	// retVal = [ max(max(x,y),z), G, G, G ]
 
 	// splat the x component out to the whole vector and return
 	return SplatXSIMD( retval );
@@ -1757,7 +1757,7 @@ FORCEINLINE fltx4 AndSIMD( const fltx4 & a, const fltx4 & b )				// a & b
 	return _mm_and_ps( a, b );
 }
 
-FORCEINLINE fltx4 AndNotSIMD( const fltx4 & a, const fltx4 & b )			// a & ~b
+FORCEINLINE fltx4 AndNotSIMD( const fltx4 & a, const fltx4 & b )			// ~a & b
 {
 	return _mm_andnot_ps( a, b );
 }
@@ -1813,7 +1813,7 @@ FORCEINLINE fltx4 ReplicateX4( float flValue )
 FORCEINLINE float SubFloat( const fltx4 & a, int idx )
 {
 	// NOTE: if the output goes into a register, this causes a Load-Hit-Store stall (don't mix fpu/vpu math!)
-#if !defined _LINUX && !defined __APPLE__
+#ifndef POSIX
 	return a.m128_f32[ idx ];
 #else
 	return (reinterpret_cast<float const *>(&a))[idx];
@@ -1822,7 +1822,7 @@ FORCEINLINE float SubFloat( const fltx4 & a, int idx )
 
 FORCEINLINE float & SubFloat( fltx4 & a, int idx )
 {
-#if !defined _LINUX && !defined __APPLE__
+#ifndef POSIX
 	return a.m128_f32[ idx ];
 #else
 	return (reinterpret_cast<float *>(&a))[idx];
@@ -1836,7 +1836,7 @@ FORCEINLINE uint32 SubFloatConvertToInt( const fltx4 & a, int idx )
 
 FORCEINLINE uint32 SubInt( const fltx4 & a, int idx )
 {
-#if !defined _LINUX && !defined __APPLE__
+#ifndef POSIX
 	return a.m128_u32[idx];
 #else
 	return (reinterpret_cast<uint32 const *>(&a))[idx];
@@ -1845,7 +1845,7 @@ FORCEINLINE uint32 SubInt( const fltx4 & a, int idx )
 
 FORCEINLINE uint32 & SubInt( fltx4 & a, int idx )
 {
-#if !defined _LINUX && !defined __APPLE__
+#ifndef POSIX
 	return a.m128_u32[idx];
 #else
 	return (reinterpret_cast<uint32 *>(&a))[idx];
@@ -2120,12 +2120,12 @@ FORCEINLINE fltx4 CmpInBoundsSIMD( const fltx4 & a, const fltx4 & b )		// (a <=
 	return AndSIMD( CmpLeSIMD(a,b), CmpGeSIMD(a, NegSIMD(b)) );
 }
 
-FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b )				// MIN(a,b)
+FORCEINLINE fltx4 MinSIMD( const fltx4 & a, const fltx4 & b )				// min(a,b)
 {
 	return _mm_min_ps( a, b );
 }
 
-FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b )				// MAX(a,b)
+FORCEINLINE fltx4 MaxSIMD( const fltx4 & a, const fltx4 & b )				// max(a,b)
 {
 	return _mm_max_ps( a, b );
 }
@@ -2271,11 +2271,11 @@ FORCEINLINE fltx4 FindLowestSIMD3( const fltx4 &a )
 	fltx4 compareOne = RotateLeft( a );
 	// compareOne is [y,z,G,x]
 	fltx4 retval = MinSIMD( a, compareOne );
-	// retVal is [MIN(x,y), ... ]
+	// retVal is [min(x,y), ... ]
 	compareOne = RotateLeft2( a );
 	// compareOne is [z, G, x, y]
 	retval = MinSIMD( retval, compareOne );
-	// retVal = [ MIN(MIN(x,y),z)..]
+	// retVal = [ min(min(x,y),z)..]
 	// splat the x component out to the whole vector and return
 	return SplatXSIMD( retval );
 	
@@ -2288,11 +2288,11 @@ FORCEINLINE fltx4 FindHighestSIMD3( const fltx4 &a )
 	fltx4 compareOne = RotateLeft( a );
 	// compareOne is [y,z,G,x]
 	fltx4 retval = MaxSIMD( a, compareOne );
-	// retVal is [MAX(x,y), ... ]
+	// retVal is [max(x,y), ... ]
 	compareOne = RotateLeft2( a );
 	// compareOne is [z, G, x, y]
 	retval = MaxSIMD( retval, compareOne );
-	// retVal = [ MAX(MAX(x,y),z)..]
+	// retVal = [ max(max(x,y),z)..]
 	// splat the x component out to the whole vector and return
 	return SplatXSIMD( retval );
 	
@@ -2407,6 +2407,12 @@ FORCEINLINE i32x4 IntShiftLeftWordSIMD(const i32x4 &vSrcA, const i32x4 &vSrcB)
 // like this.
 FORCEINLINE void ConvertStoreAsIntsSIMD(intx4 * RESTRICT pDest, const fltx4 &vSrc)
 {
+#ifdef COMPILER_MSVC64
+	(*pDest)[0] = SubFloat(vSrc, 0);
+	(*pDest)[1] = SubFloat(vSrc, 1);
+	(*pDest)[2] = SubFloat(vSrc, 2);
+	(*pDest)[3] = SubFloat(vSrc, 3);
+#else
 	__m64 bottom = _mm_cvttps_pi32( vSrc );
 	__m64 top    = _mm_cvttps_pi32( _mm_movehl_ps(vSrc,vSrc) );
 
@@ -2414,6 +2420,7 @@ FORCEINLINE void ConvertStoreAsIntsSIMD(intx4 * RESTRICT pDest, const fltx4 &vSr
 	*reinterpret_cast<__m64 *>(&(*pDest)[2]) = top;
 
 	_mm_empty();
+#endif
 }
 
 

public/mathlib/ssequaternion.h

@@ -233,7 +233,7 @@ FORCEINLINE fltx4 QuaternionScaleSIMD( const fltx4 &p, float t )
 	// FIXME: nick, this isn't overly sensitive to accuracy, and it may be faster to 
 	// use the cos part (w) of the quaternion (sin(omega)*N,cos(omega)) to figure the new scale.
 	float sinom = sqrt( SubFloat( p, 0 ) * SubFloat( p, 0 ) + SubFloat( p, 1 ) * SubFloat( p, 1 ) + SubFloat( p, 2 ) * SubFloat( p, 2 ) );
-	sinom = MIN( sinom, 1.f );
+	sinom = min( sinom, 1.f );
 
 	float sinsom = sin( asin( sinom ) * t );
 

public/mathlib/vector.h

@@ -1,4 +1,4 @@
-//====== Copyright © 1996-2005, Valve Corporation, All rights reserved. =======//
+//====== Copyright © 1996-2005, Valve Corporation, All rights reserved. =======//
 //
 // Purpose: 
 //
@@ -31,6 +31,7 @@
 #include "tier0/threadtools.h"
 #include "mathlib/vector2d.h"
 #include "mathlib/math_pfns.h"
+#include "minmax.h"
 
 // Uncomment this to add extra Asserts to check for NANs, uninitialized vecs, etc.
 //#define VECTOR_PARANOIA	1
@@ -48,7 +49,11 @@
 #ifdef VECTOR_PARANOIA
 #define CHECK_VALID( _v)	Assert( (_v).IsValid() )
 #else
+#ifdef GNUC
 #define CHECK_VALID( _v)
+#else
+#define CHECK_VALID( _v)	0
+#endif
 #endif
 
 #define VecToString(v)	(static_cast<const char *>(CFmtStr("(%f, %f, %f)", (v).x, (v).y, (v).z))) // ** Note: this generates a temporary, don't hold reference!
@@ -67,7 +72,7 @@ public:
 	// Construction/destruction:
 	Vector(void); 
 	Vector(vec_t X, vec_t Y, vec_t Z);
-	Vector(vec_t XYZ); // TODO (Ilya): is this potentially a bad idea?
+	explicit Vector(vec_t XYZ); ///< broadcast initialize
 
 	// Initialization
 	void Init(vec_t ix=0.0f, vec_t iy=0.0f, vec_t iz=0.0f);
@@ -129,6 +134,7 @@ public:
 	}
 
 	vec_t	NormalizeInPlace();
+	Vector	Normalized() const;
 	bool	IsLengthGreaterThan( float val ) const;
 	bool	IsLengthLessThan( float val ) const;
 
@@ -202,9 +208,11 @@ private:
 #endif
 };
 
+FORCEINLINE void NetworkVarConstruct( Vector &v ) { v.Zero(); }
 
-
-#define USE_M64S ( ( !defined( _X360 ) ) && ( ! defined( _LINUX) ) )
+#if ( ( !defined( _X360 ) ) && ( ! defined( _LINUX) ) )
+    #define USE_M64S 1
+#endif
 
 
 
@@ -259,7 +267,7 @@ private:
 	// No assignment operators either...
 //	ShortVector& operator=( ShortVector const& src );
 
-};
+} ALIGN8_POST;
 
 
 
@@ -395,7 +403,7 @@ public:
 	
 #endif
 	float w;	// this space is used anyway
-};
+} ALIGN16_POST;
 
 //-----------------------------------------------------------------------------
 // Vector related operations
@@ -415,7 +423,9 @@ FORCEINLINE void VectorMultiply( const Vector& a, const Vector& b, Vector& resul
 FORCEINLINE void VectorDivide( const Vector& a, vec_t b, Vector& result );
 FORCEINLINE void VectorDivide( const Vector& a, const Vector& b, Vector& result );
 inline void VectorScale ( const Vector& in, vec_t scale, Vector& result );
-inline void VectorMA( const Vector& start, float scale, const Vector& direction, Vector& dest );
+// Don't mark this as inline in its function declaration. That's only necessary on its
+// definition, and 'inline' here leads to gcc warnings.
+void VectorMA( const Vector& start, float scale, const Vector& direction, Vector& dest );
 
 // Vector equality with tolerance
 bool VectorsAreEqual( const Vector& src1, const Vector& src2, float tolerance = 0.0f );
@@ -442,6 +452,31 @@ void VectorMax( const Vector &a, const Vector &b, Vector &result );
 
 // Linearly interpolate between two vectors
 void VectorLerp(const Vector& src1, const Vector& src2, vec_t t, Vector& dest );
+Vector VectorLerp(const Vector& src1, const Vector& src2, vec_t t );
+
+FORCEINLINE Vector ReplicateToVector( float x )
+{
+	return Vector( x, x, x );
+}
+
+// check if a point is in the field of a view of an object. supports up to 180 degree fov.
+FORCEINLINE bool PointWithinViewAngle( Vector const &vecSrcPosition, 
+									   Vector const &vecTargetPosition, 
+									   Vector const &vecLookDirection, float flCosHalfFOV )
+{
+	Vector vecDelta = vecTargetPosition - vecSrcPosition;
+	float cosDiff = DotProduct( vecLookDirection, vecDelta );
+
+	if ( cosDiff < 0 ) 
+		return false;
+
+	float flLen2 = vecDelta.LengthSqr();
+
+	// a/sqrt(b) > c  == a^2 > b * c ^2
+	return ( cosDiff * cosDiff > flLen2 * flCosHalfFOV * flCosHalfFOV );
+	
+}
+
 
 #ifndef VECTOR_NO_SLOW_OPERATIONS
 
@@ -453,6 +488,10 @@ Vector RandomVector( vec_t minVal, vec_t maxVal );
 
 #endif
 
+float RandomVectorInUnitSphere( Vector *pVector );
+float RandomVectorInUnitCircle( Vector2D *pVector );
+
+
 //-----------------------------------------------------------------------------
 //
 // Inlined Vector methods
@@ -516,9 +555,9 @@ inline void Vector::Init( vec_t ix, vec_t iy, vec_t iz )
 
 inline void Vector::Random( vec_t minVal, vec_t maxVal )
 {
-	x = minVal + ((float)rand() / RAND_MAX) * (maxVal - minVal);
-	y = minVal + ((float)rand() / RAND_MAX) * (maxVal - minVal);
-	z = minVal + ((float)rand() / RAND_MAX) * (maxVal - minVal);
+	x = minVal + ((float)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
+	y = minVal + ((float)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
+	z = minVal + ((float)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
 	CHECK_VALID(*this);
 }
 
@@ -1081,14 +1120,6 @@ inline void VectorScale ( const Vector& in, vec_t scale, Vector& result )
 	VectorMultiply( in, scale, result );
 }
 
-inline void VectorMA( const Vector& start, float scale, const Vector& direction, Vector& dest )
-{
-	CHECK_VALID(start);
-	CHECK_VALID(direction);
-	dest.x = start.x + scale * direction.x;
-	dest.y = start.y + scale * direction.y;
-	dest.z = start.z + scale * direction.z;
-}
 
 FORCEINLINE void VectorDivide( const Vector& a, vec_t b, Vector& c )
 {
@@ -1130,6 +1161,12 @@ inline void VectorLerp(const Vector& src1, const Vector& src2, vec_t t, Vector&
 	dest.z = src1.z + (src2.z - src1.z) * t;
 }
 
+inline Vector VectorLerp(const Vector& src1, const Vector& src2, vec_t t )
+{
+	Vector result;
+	VectorLerp( src1, src2, t, result );
+	return result;
+}
 
 //-----------------------------------------------------------------------------
 // Temporary storage for vector results so const Vector& results can be returned
@@ -1430,6 +1467,13 @@ inline void VectorMax( const Vector &a, const Vector &b, Vector &result )
 	result.z = fpmax(a.z, b.z);
 }
 
+inline float ComputeVolume( const Vector &vecMins, const Vector &vecMaxs )
+{
+	Vector vecDelta;
+	VectorSubtract( vecMaxs, vecMins, vecDelta );
+	return DotProduct( vecDelta, vecDelta );
+}
+
 // Get a random vector.
 inline Vector RandomVector( float minVal, float maxVal )
 {
@@ -1609,7 +1653,7 @@ public:
 	}
 
 #endif
-};
+} ALIGN16_POST;
 
 
 //-----------------------------------------------------------------------------
@@ -1642,6 +1686,9 @@ public:
 
 extern void AngleQuaternion( RadianEuler const &angles, Quaternion &qt );
 extern void QuaternionAngles( Quaternion const &q, RadianEuler &angles );
+
+FORCEINLINE void NetworkVarConstruct( Quaternion &q ) { q.x = q.y = q.z = q.w = 0.0f; }
+
 inline Quaternion::Quaternion(RadianEuler const &angle)
 {
 	AngleQuaternion( angle, *this );
@@ -1789,6 +1836,8 @@ private:
 #endif
 };
 
+FORCEINLINE void NetworkVarConstruct( QAngle &q ) { q.x = q.y = q.z = 0.0f; }
+
 //-----------------------------------------------------------------------------
 // Allows us to specifically pass the vector by value when we need to
 //-----------------------------------------------------------------------------
@@ -1852,9 +1901,9 @@ inline void QAngle::Init( vec_t ix, vec_t iy, vec_t iz )
 
 inline void QAngle::Random( vec_t minVal, vec_t maxVal )
 {
-	x = minVal + ((float)rand() / RAND_MAX) * (maxVal - minVal);
-	y = minVal + ((float)rand() / RAND_MAX) * (maxVal - minVal);
-	z = minVal + ((float)rand() / RAND_MAX) * (maxVal - minVal);
+	x = minVal + ((float)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
+	y = minVal + ((float)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
+	z = minVal + ((float)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
 	CHECK_VALID(*this);
 }
 
@@ -2127,11 +2176,16 @@ inline void AngularImpulseToQAngle( const AngularImpulse &impulse, QAngle &angle
 }
 
 #if !defined( _X360 )
-extern float (*pfInvRSquared)( const float *v );
 
 FORCEINLINE vec_t InvRSquared( float const *v )
 {
-	return (*pfInvRSquared)(v);
+#if defined(__i386__) || defined(_M_IX86)
+	float sqrlen = v[0]*v[0]+v[1]*v[1]+v[2]*v[2] + 1.0e-10f, result;
+	_mm_store_ss(&result, _mm_rcp_ss( _mm_max_ss( _mm_set_ss(1.0f), _mm_load_ss(&sqrlen) ) ));
+	return result;
+#else
+	return 1.f/fpmax(1.f, v[0]*v[0]+v[1]*v[1]+v[2]*v[2]);
+#endif
 }
 
 FORCEINLINE vec_t InvRSquared( const Vector &v )
@@ -2139,36 +2193,63 @@ FORCEINLINE vec_t InvRSquared( const Vector &v )
 	return InvRSquared(&v.x);
 }
 
-#else
-
-// call directly
-FORCEINLINE float _VMX_InvRSquared( const Vector &v )
+#if defined(__i386__) || defined(_M_IX86)
+inline void _SSE_RSqrtInline( float a, float* out )
 {
-	XMVECTOR xmV = XMVector3ReciprocalLength( XMLoadVector3( v.Base() ) );
-	xmV = XMVector3Dot( xmV, xmV );
-	return xmV.x;
+	__m128  xx = _mm_load_ss( &a );
+	__m128  xr = _mm_rsqrt_ss( xx );
+	__m128  xt;
+	xt = _mm_mul_ss( xr, xr );
+	xt = _mm_mul_ss( xt, xx );
+	xt = _mm_sub_ss( _mm_set_ss(3.f), xt );
+	xt = _mm_mul_ss( xt, _mm_set_ss(0.5f) );
+	xr = _mm_mul_ss( xr, xt );
+	_mm_store_ss( out, xr );
 }
-
-#define InvRSquared(x) _VMX_InvRSquared(x)
-
-#endif // _X360
-
-#if !defined( _X360 )
-extern float (FASTCALL *pfVectorNormalize)(Vector& v);
+#endif
 
 // FIXME: Change this back to a #define once we get rid of the vec_t version
-FORCEINLINE float VectorNormalize( Vector& v )
+FORCEINLINE float VectorNormalize( Vector& vec )
 {
-	return (*pfVectorNormalize)(v);
+#ifndef DEBUG // stop crashing my edit-and-continue!
+	#if defined(__i386__) || defined(_M_IX86)
+		#define DO_SSE_OPTIMIZATION
+	#endif
+#endif
+
+#if defined( DO_SSE_OPTIMIZATION )
+	float sqrlen = vec.LengthSqr() + 1.0e-10f, invlen;
+	_SSE_RSqrtInline(sqrlen, &invlen);
+	vec.x *= invlen;
+	vec.y *= invlen;
+	vec.z *= invlen;
+	return sqrlen * invlen;
+#else
+	extern float (FASTCALL *pfVectorNormalize)(Vector& v);
+	return (*pfVectorNormalize)(vec);
+#endif
 }
+
 // FIXME: Obsolete version of VectorNormalize, once we remove all the friggin float*s
 FORCEINLINE float VectorNormalize( float * v )
 {
 	return VectorNormalize(*(reinterpret_cast<Vector *>(v)));
 }
 
+FORCEINLINE void VectorNormalizeFast( Vector &vec )
+{
+	VectorNormalize(vec);
+}
+
 #else
 
+FORCEINLINE float _VMX_InvRSquared( const Vector &v )
+{
+	XMVECTOR xmV = XMVector3ReciprocalLength( XMLoadVector3( v.Base() ) );
+	xmV = XMVector3Dot( xmV, xmV );
+	return xmV.x;
+}
+
 // call directly
 FORCEINLINE float _VMX_VectorNormalize( Vector &vec )
 {
@@ -2179,6 +2260,9 @@ FORCEINLINE float _VMX_VectorNormalize( Vector &vec )
 	vec.z *= den;
 	return mag;
 }
+
+#define InvRSquared(x) _VMX_InvRSquared(x)
+
 // FIXME: Change this back to a #define once we get rid of the vec_t version
 FORCEINLINE float VectorNormalize( Vector& v )
 {
@@ -2190,18 +2274,6 @@ FORCEINLINE float VectorNormalize( float *pV )
 	return _VMX_VectorNormalize(*(reinterpret_cast<Vector*>(pV)));
 }
 
-#endif // _X360
-
-#if !defined( _X360 )
-extern void (FASTCALL *pfVectorNormalizeFast)(Vector& v);
-
-FORCEINLINE void VectorNormalizeFast( Vector& v )
-{
-	(*pfVectorNormalizeFast)(v);
-}
-
-#else
-
 // call directly
 FORCEINLINE void VectorNormalizeFast( Vector &vec )
 {
@@ -2214,11 +2286,19 @@ FORCEINLINE void VectorNormalizeFast( Vector &vec )
 
 #endif // _X360
 
+
 inline vec_t Vector::NormalizeInPlace()
 {
 	return VectorNormalize( *this );
 }
 
+inline Vector Vector::Normalized() const
+{
+	Vector norm = *this;
+	VectorNormalize( norm );
+	return norm;
+}
+
 inline bool Vector::IsLengthGreaterThan( float val ) const
 {
 	return LengthSqr() > val*val;

public/mathlib/vector2d.h

@@ -1,4 +1,4 @@
-//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
+//========= Copyright © 1996-2005, Valve Corporation, All rights reserved. ============//
 //
 // Purpose: 
 //
@@ -239,8 +239,8 @@ inline void Vector2D::Init( vec_t ix, vec_t iy )
 
 inline void Vector2D::Random( float minVal, float maxVal )
 {
-	x = minVal + ((float)rand() / RAND_MAX) * (maxVal - minVal);
-	y = minVal + ((float)rand() / RAND_MAX) * (maxVal - minVal);
+	x = minVal + ((float)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
+	y = minVal + ((float)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
 }
 
 inline void Vector2DClear( Vector2D& a )

public/mathlib/vector4d.h

@@ -132,11 +132,7 @@ const Vector4D vec4_invalid( FLT_MAX, FLT_MAX, FLT_MAX, FLT_MAX );
 // SSE optimized routines
 //-----------------------------------------------------------------------------
 
-#ifdef _WIN32
-class __declspec(align(16)) Vector4DAligned : public Vector4D
-#elif defined _LINUX || defined __APPLE__
-class __attribute__((aligned(16))) Vector4DAligned : public Vector4D
-#endif
+class ALIGN16 Vector4DAligned : public Vector4D
 {
 public:
 	Vector4DAligned(void) {}
@@ -154,7 +150,7 @@ private:
 
 	// No assignment operators either...
 	Vector4DAligned& operator=( Vector4DAligned const& src );
-};
+} ALIGN16_POST;
 
 //-----------------------------------------------------------------------------
 // Vector4D related operations
@@ -249,10 +245,10 @@ inline void Vector4D::Init( vec_t ix, vec_t iy, vec_t iz, vec_t iw )
 
 inline void Vector4D::Random( vec_t minVal, vec_t maxVal )
 {
-	x = minVal + ((vec_t)rand() / RAND_MAX) * (maxVal - minVal);
-	y = minVal + ((vec_t)rand() / RAND_MAX) * (maxVal - minVal);
-	z = minVal + ((vec_t)rand() / RAND_MAX) * (maxVal - minVal);
-	w = minVal + ((vec_t)rand() / RAND_MAX) * (maxVal - minVal);
+	x = minVal + ((vec_t)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
+	y = minVal + ((vec_t)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
+	z = minVal + ((vec_t)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
+	w = minVal + ((vec_t)rand() / VALVE_RAND_MAX) * (maxVal - minVal);
 }
 
 inline void Vector4DClear( Vector4D& a )

public/mathlib/vmatrix.h

@@ -54,6 +54,7 @@ public:
 	// Creates a matrix where the X axis = forward
 	// the Y axis = left, and the Z axis = up
 	VMatrix( const Vector& forward, const Vector& left, const Vector& up );
+	VMatrix( const Vector& forward, const Vector& left, const Vector& up, const Vector& translation );
 	
 	// Construct from a 3x4 matrix
 	VMatrix( const matrix3x4_t& matrix3x4 );
@@ -106,7 +107,6 @@ public:
 	void		PreTranslate(const Vector &vTrans);
 	void		PostTranslate(const Vector &vTrans);
 
-	matrix3x4_t& As3x4();
 	const matrix3x4_t& As3x4() const;
 	void		CopyFrom3x4( const matrix3x4_t &m3x4 );
 	void		Set3x4( matrix3x4_t& matrix3x4 ) const;
@@ -199,6 +199,9 @@ public:
 	// Setup a matrix for origin and angles.
 	void		SetupMatrixOrgAngles( const Vector &origin, const QAngle &vAngles );
 	
+	// Setup a matrix for angles and no translation.
+	void		SetupMatrixAngles( const QAngle &vAngles );
+
 	// General inverse. This may fail so check the return!
 	bool		InverseGeneral(VMatrix &vInverse) const;
 	
@@ -457,6 +460,16 @@ inline VMatrix::VMatrix( const Vector& xAxis, const Vector& yAxis, const Vector&
 		);
 }
 
+inline VMatrix::VMatrix( const Vector& xAxis, const Vector& yAxis, const Vector& zAxis, const Vector& translation )
+{
+	Init(
+		xAxis.x, yAxis.x, zAxis.x, translation.x,
+		xAxis.y, yAxis.y, zAxis.y, translation.y,
+		xAxis.z, yAxis.z, zAxis.z, translation.z,
+		0.0f, 0.0f, 0.0f, 1.0f
+		);
+}
+
 
 inline void VMatrix::Init(
 	vec_t m00, vec_t m01, vec_t m02, vec_t m03,
@@ -616,11 +629,6 @@ inline const matrix3x4_t& VMatrix::As3x4() const
 	return *((const matrix3x4_t*)this);
 }
 
-inline matrix3x4_t& VMatrix::As3x4()
-{
-	return *((matrix3x4_t*)this);
-}
-
 inline void VMatrix::CopyFrom3x4( const matrix3x4_t &m3x4 )
 {
 	memcpy( m, m3x4.Base(), sizeof( matrix3x4_t ) );
@@ -902,6 +910,7 @@ inline bool MatricesAreEqual( const VMatrix &src1, const VMatrix &src2, float fl
 void MatrixBuildOrtho( VMatrix& dst, double left, double top, double right, double bottom, double zNear, double zFar );
 void MatrixBuildPerspectiveX( VMatrix& dst, double flFovX, double flAspect, double flZNear, double flZFar );
 void MatrixBuildPerspectiveOffCenterX( VMatrix& dst, double flFovX, double flAspect, double flZNear, double flZFar, double bottom, double top, double left, double right );
+void MatrixBuildPerspectiveZRange( VMatrix& dst, double flZNear, double flZFar );
 
 inline void MatrixOrtho( VMatrix& dst, double left, double top, double right, double bottom, double zNear, double zFar )
 {

public/mdllib/mdllib.h

@@ -72,7 +72,7 @@ public:
 	// Returns the checksums that the stripping info was generated for:
 	//	plChecksumOriginal		if non-NULL will hold the checksum of the original model submitted for stripping
 	//	plChecksumStripped		if non-NULL will hold the resulting checksum of the stripped model
-	virtual bool GetCheckSum( long *plChecksumOriginal, long *plChecksumStripped ) const = 0;
+	virtual bool GetCheckSum( int32_t *plChecksumOriginal, int32_t *plChecksumStripped ) const = 0;
 
 	//
 	// Stripping