havoc [Fri, 13 May 2005 03:24:13 +0000 (03:24 +0000)]
finished model/map rendering merge, model renderer has been completely removed
model_t num_textures now lies a bit to keep things simple, it equals num_surfaces on models, even if there are multiple skins, in data_textures these extra textures really do exist even though num_textures does not count them (this was done so that UpdateAllTextureInfo won't crash trying to index later skins on the later skins themselves)
havoc [Thu, 12 May 2005 20:44:12 +0000 (20:44 +0000)]
labeled todo items (at least ones not done yet) with a category before the program name, to prioritize bugfixes over changes and features and so on within a difficulty rating
cleaned out a lot of stale or duplicate todo items
376 (-20)
black [Wed, 11 May 2005 20:26:12 +0000 (20:26 +0000)]
-Changed VM_precache_pic to check CachePic for successful loading of the pictures.
(The old check hasnt worked because CachePic always returns a valid pic pointer.)
havoc [Wed, 11 May 2005 18:22:48 +0000 (18:22 +0000)]
the merging of alias and bsp model rendering begins... purely cosmetic changes in this first step
merged model_alias_t into model_t and removed alias prefixes from those fields
merged model_brush_t surface and texture stuff into model_t (num_surfaces, num_textures, etc)
made alias model code use model_t data_surfaces instead of alias meshes
havoc [Mon, 9 May 2005 13:03:18 +0000 (13:03 +0000)]
fixed problems in the Host_Init Cbuf_InsertText stuff (by changing it to Cbuf_AddText, and some other cleanups)
added Host_StartVideo to a few more places
havoc [Sun, 8 May 2005 19:57:24 +0000 (19:57 +0000)]
implemented caching of DNS names in lhnet.c
changed sv_masterextra* servers back to DNS names (but kept IP addresses as comments, note the dpmaster.deathmask.net IP has changed)
havoc [Sun, 8 May 2005 13:22:50 +0000 (13:22 +0000)]
revised a number of terminology notes at the beginning of r_shadow.c, including notes about the Creative Labs patent on the Carmack's Reverse technique
havoc [Sun, 8 May 2005 12:59:38 +0000 (12:59 +0000)]
added rmesh_t and R_Mesh_AddVertex3f, R_Mesh_AddPolygon3f, R_Mesh_AddBrushMeshFromPlanes functions
R_Shadow_ScissorForBBox now uses R_Mesh_AddBrushMeshFromPlanes function to get an exact clipped box-brush vertex set for calculating scissor area, this should reduce the scissor area used in some cases compared to the old hacky method
added nearclip plane to frustum[] array (only used by R_Shadow_ScissorForBBox)
havoc [Fri, 6 May 2005 11:37:35 +0000 (11:37 +0000)]
changed Host_Init to execute configs only once
cleaned up Host_Init (and related functions like Con_Init, COM_Init, Cbuf_Init, etc) a great deal
now uses host_framecount >= 3 checks instead of host_loopactive or host_initialized checks, this should fix any problems with crashes erasing config.cfg
VID_Open and friends are now called from Host_StartVideo which is called by SCR_BeginLoadingPlaque (such as by SV_SpawnServer) as well as the end of Host_Init, this only calls them the first time it is called, so it can be called during config parsing (such as +map start on the commandline)
moved sys_usetimegettime and dedicated server console opening in sys_wgl.c into Sys_InitConsole and Sys_Init_Commands functions called by Host_Init
merged Sys_Shared_EarlyInit and Sys_Shared_LateInit into Host_Init
got rid of sys_usetimegettime (windows-only) cvar in sys_sdl.c
changed type of a number of vid_* variables from int to qboolean
renamed vidmode_active to vid_isfullscreen in vid_glx.c for more code consistency
fixed a bug in vid_glx.c that made it grab the mouse/keyboard even when it was not the active window
no longer grabs mouse while playing demos in a window
havoc [Fri, 6 May 2005 11:28:31 +0000 (11:28 +0000)]
moved mod_shared.c detail texture and distortion texture stuff to gl_rmain.c (renamed mod_shared_* to r_texture_*)
got rid of unused mod_q1bsp_novis array
moved Mod_Q1BSP_Collision_Init into the boxhull code (which isn't even used anyway)
made r_restart reload models (as was intended) by making Mod_UnloadModel preserve the ->use field
havoc [Fri, 6 May 2005 06:09:33 +0000 (06:09 +0000)]
removed a clear of names in the render modules init code, this allows it to be called AFTER some modules are registered (for a minor cleanup of the Host_Init code)
havoc [Fri, 6 May 2005 00:36:07 +0000 (00:36 +0000)]
added cl.movement_velocity to allow view bobbing to match up with cl_movement
fixed some bugs in cl_movement timing, it now applies input to the next move not the current one, making it work properly (assuming cl_movement_latency is exact)
black [Thu, 5 May 2005 12:48:43 +0000 (12:48 +0000)]
-Changed the SDL window icon back to DP's icon.
-Cmds, Cvars and Aliases are now inserted at the right alphanumerical
position on creation.
-Fixed a bug in the serverlist which caused it to not mask entries properly.
havoc [Wed, 4 May 2005 20:14:06 +0000 (20:14 +0000)]
changed cl_movement_latency to be in milliseconds rather than seconds (less confusing because of ping being milliseconds)
added code to remove 'impossibly far into the future' movement queue items (useful if latency is set to a high value initially and then lower, this will remove the stale high latency items)
relocated the CL_SendMove call to put it at the end of network parsing (and only called if there was an svc_time message in this packet), this should greatly improve consistency of ping times and input sending (which was sometimes falling way behind with lag), no longer uses sys_ticrate to control input frequency
changed CL_TraceLine to CL_TraceBox and made it return a trace_t
worked on CL_ClientMovement function for clientside movement simulation (not working yet), this has many cl_movement* cvars associated with it which will be updated by the server at some point in the future
added (currently unused and not updated) cl_gravity and cl_slowmo cvars, which the server will be setting at some point in the future
slight cleanup of surface clipping code (got rid of a rather redundent function)
major bugfix to surface portal clipping, lights should hit all the right surfaces now
fixed CL_SendMove packet rate limiting to not have a heartattack when going from singleplayer to multiplayer (this was causing the player to be stuck in place for a very long time)
added cl_capturevideo_sound (defaulted to 0) to allow enabling/disabling sound saving, with sound disabled you can save videos at a framerate your machine can't maintain (no sound sync to worry about)
reimplemented rtlight portal culling (used for compiled rtlights, and can be used for dlights but that is often a major speed loss)
added additional bounding box and triangle facing checks to rtlight compilation, triangles outside the box are no longer drawn
r_shadow_staticworldlights has mutated into r_shadow_realtime_world_compile, r_shadow_realtime_world_compilelight, r_shadow_realtime_world_compileshadow
added r_shadow_visiblelighting
fixed a stupid bug in SV_PushMove that made it ignore the first entity in the move box
fixed a rotation check in SV_PushMove so that it would detect roll rotations without pitch/yaw
gl_texture_anisotropy now only affects TEXF_MIPMAP textures which should fix a number of 'wandering off the lightmap' issues (as well as similar issues with model skins, as they are not mipmapped)
texture images now take on the TEXF_FORCENEAREST and TEXF_FORCELINEAR flags from the textures placed in them (as was intended), so gl_texturemode changes
should no longer affect them
default to insert mode instead of replace mode in console
get rid of stupid cursor right at end of line behavior in console, repeating characters from the previous edit line is annoying
allow typing characters > 126 in console
changed R_Shadow_RenderLighting to do full base+pants+shirt+gloss rendering in one call (will optimize the rendering itself later to merge some of the passes)
made dland2.wav not play in GAME_NEXUIZ
no longer runs physics on disconnected clients (quake did not run them), fixing the physics problems with frikbots, thanks to FrikaC for explanation of this problem
black [Thu, 14 Apr 2005 20:45:08 +0000 (20:45 +0000)]
-Added the parameter 'persistent'(naming?) to Draw_CachePic, which decides
whether the texture is loaded in texture memory directly or only on demand.
-Adapted all Draw_CachePic calls (I'll do some corrections, atm most of them
pass persistent := false)
LoadTGA is now more compliant with the TGA spec regarding colormaps (they can be present even on a truecolor image, just not used by it) and attributebits (alpha is now ignored if attributebits is not 8 which fixes a number of malformed TGA files (though they still waste disk space) - note that attributebits is part of the attributes header byte)
projects / xonotic / darkplaces.git / log