 |
Why am I monitoring this? Do I need to do something? |
 |
I'm just putting this out on alert as it's critical to 2.06. All hands on deck, etc. I'll remove you as a monitor. |
|
|
I believe it was this commit: https://github.com/fholger/thedarkmod/commit/9159212ca808763d14e4b9677e48f0d58b24272e but I'll have to retest old builds to be sure. The fragility here is probably stemming from the fact that the PBO "initialization" is in Rendersystem rather than Rendersystem_init. Another relevant commit. Where the lightgem render order was moved: https://github.com/fholger/thedarkmodvr/commit/ae25e4d567fe2e4f97d6425d57d3c2ee166dc61e |
|
|
Looks like we want it here: =============== idRenderSystemLocal::Clear =============== */ void idRenderSystemLocal::Clear( void ) { registered = false; frameCount = 0; viewCount = 0; staticAllocCount = 0; frameShaderTime = 0.0f; viewportOffset[0] = 0; viewportOffset[1] = 0; tiledViewport[0] = 0; tiledViewport[1] = 0; ambientLightVector.Zero(); sortOffset = 0; worlds.Clear(); primaryWorld = NULL; memset( &primaryRenderView, 0, sizeof( primaryRenderView ) ); primaryView = NULL; defaultMaterial = NULL; testImage = NULL; ambientCubeImage = NULL; viewDef = NULL; memset( &pc, 0, sizeof( pc ) ); memset( &lockSurfacesCmd, 0, sizeof( lockSurfacesCmd ) ); memset( &identitySpace, 0, sizeof( identitySpace ) ); logFile = NULL; stencilIncr = 0; stencilDecr = 0; memset( renderCrops, 0, sizeof( renderCrops ) ); currentRenderCrop = 0; guiRecursionLevel = 0; guiModel = NULL; demoGuiModel = NULL; memset( gammaTable, 0, sizeof( gammaTable ) ); takingScreenshot = false; } /* =============== idRenderSystemLocal::Init =============== */ void idRenderSystemLocal::Init( void ) { // clear all our internal state viewCount = 1; // so cleared structures never match viewCount // we used to memset tr, but now that it is a class, we can't, so // there may be other state we need to reset ambientLightVector[0] = 0.5f; ambientLightVector[1] = 0.5f - 0.385f; ambientLightVector[2] = 0.8925f; ambientLightVector[3] = 1.0f; memset( &backEnd, 0, sizeof( backEnd ) ); R_InitCvars(); R_InitCommands(); guiModel = new idGuiModel; guiModel->Clear(); demoGuiModel = new idGuiModel; demoGuiModel->Clear(); R_InitTriSurfData(); globalImages->Init(); idCinematic::InitCinematic( ); // build brightness translation tables R_SetColorMappings(); R_InitMaterials(); renderModelManager->Init(); // set the identity space identitySpace.modelMatrix[0*4+0] = 1.0f; identitySpace.modelMatrix[1*4+1] = 1.0f; identitySpace.modelMatrix[2*4+2] = 1.0f; // determine which back end we will use // FIXME : ??? this is invalid here as there is not enough information to set it up correctly //SetBackEndRenderer(); } |
|
|
Maybe get shutdown too: =============== idRenderSystemLocal::Shutdown =============== */ void idRenderSystemLocal::Shutdown( void ) { common->Printf( "idRenderSystem::Shutdown()\n" ); R_DoneFreeType( ); if ( glConfig.isInitialized ) { globalImages->PurgeAllImages(); } renderModelManager->Shutdown(); idCinematic::ShutdownCinematic( ); globalImages->Shutdown(); // close the r_logFile if ( logFile ) { fprintf( logFile, "*** CLOSING LOG ***\n" ); fclose( logFile ); logFile = 0; } // free frame memory R_ShutdownFrameData(); // free the vertex cache, which should have nothing allocated now vertexCache.Shutdown(); R_ShutdownTriSurfData(); RB_ShutdownDebugTools(); delete guiModel; delete demoGuiModel; Clear(); ShutdownOpenGL(); } |
 |
Don't do that, please. Just putting everyone on the monitor list and have them receive emails out of nowhere is sort of rude. Whoever did that commit should fix it, I definitely don't bother. I've had my share of doing other people's homework when I was actively working on TDM. |
|
|
Understood. I'll try fiddling with this tonight. Last night SVN wouldn't compile due to an uncommitted local file in "someone's" repo... I'm not the best with coding standards\practices (etc) so I was hoping for some veteran to chime in. I know orbweaver used to speak on this topic quite extensively. |
 |
What about vid_restart without mission change? It works for me. I did not try the mission change though. |
|
|
I was trying to use vid_restart as a workaround. In those cases it wouldn't cure the problem. I will test it in the scenario where the Lightgem works. |
|
|
Invoking vid_restart (after mission load on a fresh executable start with an installed mission) breaks the lightgem too. Even with safe defaults: com_smp 0 com_fixedTic 0 |
|
|
Rev 7156 Binaries 14936 |
- Anonymous
