A value of 1 in freesrc for Mix_LoadWAV_RW and Mix_LoadMus_RW calls SDL_RWclose on the RWops anyway.
For Mix_LoadWAV_RW the RWops is freed right after the data is loaded (because it makes a copy of the data in memory)
For Mix_LoadMUS_RW the RWops is freed when Mix_FreeMusic is called (because the data is not a copy)
So setting 1 on freesrc doesn't actually free the RWops immediately on Mix_LoadMus_RW *unless* it failed to load any music.
Checks for the flag when freeing, and if it's 0, we free the data manually after Mix_FreeChunk.
I went back to Z_Malloc and Z_Free for this because they still work after this.
* Mixer code has been in mixer_sound.c; this file is not invoked unless compiling with NOMIXER=1
* Remove everything under #ifdef HAVE_MIXER because this is never triggered
* Comment out #ifdef HAVE_LIBGME because we don't support playing music anyway (but theoretically, it could have worked separately from Mixer)
* Stub new music calls
* S_Init -> S_InitSfxChannels because it did mostly SFX anyway
* S_MusicPlaying, S_MusicPaused, S_MusicName, S_MusicExists new status methods
* I_MusicPlaying, I_MusicPaused
(cherry picked from commit f5f0b5e76c)
* In s_sound, they are merged to one method as well, but there are still two separate digvolume and seqvolume variables
* Simplified Dig/MidiMusicDisabled in s_sound
* Method reordering
(cherry picked from commit 701cc5a7dd)
Use > 90 && < 270 instead of >= 90 && <= 270. Fixes a bug where if you fly directly up (flyangle 90) or directly down (flyangle 270), that registers as a backwards direction, so you trigger the loop detection by flying BACKWARDS, not FORWARDS. This edge case (only possible via JUMPTOAXIS) should default to FORWARDS looping.
(cherry picked from commit ce215195f8)
With that, I moved R_CreateColormap2's exclusive software colormap malloc code to R_CreateColormap, and merged the two software-only blocks of code into one. I also disabled any unneeded variables and fixed a preprocessor-related goofup
* V_DrawFixedPatch and ilk:
* Change the offset of V_FLIP so it's not one screen-pixel off where its non-flipped sprite would have started being drawn from.
* Write to x and y as well as desttop so that anti-screen-overflow checks later in the function behave properly with non-green resolutions.
* V_DrawFill:
* Reduce number of operations performed upon `c`.
* V_DrawString and ilk:
* Offset the left and right boundary checks in non-green resolutions such that you can actually draw stuff to the left of basevid screen x coordinate 0.
* Stop orphaning their memory. They ARE PU_LEVEL, so they'll disappear eventually, but, like... it's not good memory management practice to just *orphan* them when you're literally never going to do anything with them ever again. Y'know?
* Make ghosts spawn properly on slopes.
* Fix that thing where ALL transparent FOF planes were continuously fullbright unless encased in a fog which disables sprite fullbrightness, which was long-hated by many people in the community!
* For backwards compatibility, setting flag 1 in that fog field (which is probably the most common "in-the-wild" usage of this feature) will continue to make objects un-fullbright.
* For situations where you desperately want the behaviour to be enabled, you can apply fog flag 2.
* Change the fadestart and fadeend range in which colormaps are generated.
* The problem HERE was that the darkest light level reached by generated colormaps was actually slightly brighter than the darkest level reached by normal colormaps.
* The typo I fixed does have SOME basis in fact - standard colormap lumps are 34 (33 in 0-indexing) long rather than 32 (31), but whoever wrote this didn't realise that the code for generating them didn't do it DooM style, just bright-to-dark with no extras on the end...
This now means:
* Lua strings longer than 1024 chars can now be read properly without awful crashes
* Lua strings with embedded zeros can be written/read without truncating anything (hopefully)
Also tweaked a weird splitscreen check in HWR_DrawSpriteShadow; still investigating whether stplyr is ever not player 2 when it's player 2's view, but this looks better for now
Move old fix for too large maps having rendering issues from R_CheckBBox to OpenGL's HWR_CheckBBox
From what I know, this effects at least Aerial Garden and Seraphic Skylands
The FixedMul() C implementation would produce off by one results,
causing constant desyncs on 64 bit builds and builds without an
ASM implementation of the function.
This is fixed by shifting instead of dividing, possibly avoiding
rounding errors.
gcc accepts
__asm__ ( "" : : : "%cc");
but not clang:
error: unknown register name '%cc' in asm
:"%cc", "%edx" // edx and condition codes clobbered
'cc' is not a register, thus '%cc' is rightfully not accepted.
Repro: build the project on x86_64 with:
cmake -H. -Bbuild -DCMAKE_C_COMPILER=clang -DCMAKE_SIZEOF_VOID_P=4 -DCMAKE_C_FLAGS="-m32"
This makes OpenGL stop using a specific function that doesn't really do anything for it anymore. It looks like it was used for a hack that would change the colour of polygons for the flashpal equivalent in DOOM.
I made it so ST_DoPaletteStuff doesn't set the flashpal in OpenGL as it already does its own hacky overlay and doing that would cause all the textures to be flushed more mid-level, it could be enabled for more correct flashpals, but they still wouldn't effect fog or lighting.
This means the palette will be set when going to the title screen, and twice when starting a map, (causing the OpenGL cached textures to also be flushed at those times)
Also move the Bouncy FOF sector special check above the FOF heights checking in P_CheckBouncySectors, because it means not having to waste time calculating FOF heights only for it not to be bouncy anyway :P
This means that, if the three paths are not the same, you should be able to tell if at least one of them has a file that just had a bad MD5. Most relevant for Linux peeps I expect.
Note: Untested as of writing
I don't fully understand this, but it's what software does and it fixes the issue of the lighting in DSZ3. Also don't need the extra call to R_Prep3DFloors.
I don't think it does anything for us anymore, and might even break things with slopes.
Someone let me know if I'm wrong and am breaking things horribly here.
Transformation based on screen space would make sense if we didn't want anything in the world to effect the sprites.
This should allow sprite splitting and sorting of sprites with level geometry easier.
stransform is no longer needed.
Solid walls *can* be cut
Fix issues with water and fog FOFs not cutting each other out correctly
Fix Fog colourmap and lighting setting that is done here.
Remove HWR_SplitFog
There is currently a bug with FF_DOUBLESHADOW (that also exists in software) but has a larger impact here. When 2 FF_DOUBLESHADOW lights are directly stacked on each other the bottom one has its height set incorrectly. This causes all the Fog in the timed gravity flipping section of ERZ2 to be drawn and it looks really bad.
This reverts commit 121fcd8369.
The reason I am reverting this is because the last commit actually fixes the *old* screenshot functionality, as the screen is being drawn back onto the buffer after they're swapped in the "real" size. Meaning the old function actually works perfectly fine now.
They still aren't perfect, but now they are at least not quite so obviously just translucent polygons over the level. A mixture between partially modulating the background colours and adding the fog colour. Notably white fog blocks look like they're brightening what's behind them.
Additive was also setting noalphatest before, can probably decide that depending on what it needs anyway. I don't think it's currently used anyway.
Sorts all translucent sprites and MD2s so they're drawn after all the opaque ones. Fixes most of the observable issues between translucent MD2s and opaque sprites/MD2s.
Only use glCopyTexImage2D when first creating the screen texture, use glCopyTexSubImage2D anytime after that as it does not define a new texture each time.
Flushing of the screen textures has been implemented for when the screen size changes (so that the screen textures don't stay at a wrong size) and the game is closed, I believe they would leave a memory leak before.
The Far clipping plane did not need to be nearly as high as it was, the new value is 32768, which I suspect is about how far software can render before it completely falls apart.
It is desirable to increase the near clipping plane to between 6-10, but it can introduce more issues with close geometry not being drawn when the player or camera is scaled or viewheight is set to MIN in first person view. It would also stop sprites from being drawn ever so slightly too early, but this isn't too much of an issue and isn't too noticeable with those values. Might look into scaling near clipping plane in accordance to camera scale in the future.
The reason for wanting to increase the near clipping plane is because the small value can cause very noticeable Z-fighting where there shouldn't be on older GPU's, usually Intel ones, that don't support 24-bits for the depth buffer.
* fix screen to properly truncate the filename to just the real name only
* if the real name itself is too long, use ellipsis and paste in parts of the start and end of the actual name
note: I haven't actually tested if this works or compiles yet, I haven't the time right now
Turns out compiling for Linux 32-bit using the Makefiles never actually defines LINUX! Apart from that, most of the existing Linux-specific code in SRB2's source code (except for tmap.s) actually uses __linux__ instead anyway
* ArchiveTables: print an error if invalid key, to alert script author potentially
* UnArchiveTables: if the key is found to be nil after reading key and value, print an error and don't set them in the table
Skewing direction is decided per in-level wall by the Lower Unpegged flag on in-level linedefs themselves, since they already decide the stuff for FOF wall pegging as it is. That is unless Transfer Line is involved which moves everything to the control sector linedefs instead...
This prevents a mobj of NUMMOBJTYPES from being created and potentially crashing the game. Really, this happens only if you start messing with the mapthings of respawning items like rings/weapons/etc in the first place via Lua.
Also removed the + 1 from newtime, since there was never really any need for it. It just offset the interpolation so it went like (1 -> 2] instead of [1 -> 2), so you never saw the base appearance for each frame except at the end of any frames interpolating to it
Changed DrawMD2Ex's duration/tics type to INT32 so -1 comparisons work, probably want to change the signs elsewhere too but this is fine for now
Draw Textures and Flats that have holes in them like a solid polygon so they use the depth buffer and don't need to be sorted
Disable all linear filtering on textures and flats that have holes in them, the linear filtering introduces translucency into the textures where the edges are. Leaving them with either a black border, or causing pixels behind the slightly translucent areas to not be drawn. Doesn't apply to sprites and the HUD as they are always already sorted properly.
Make the Alpha Testing more strict on non-translucent blend modes. This makes it so any transparency below 0.5 is discarded instead. Would make anything that is blended and has holes in it look slightly better, only the HUD and MD2s where the texture has holes are effected currently.
Set TF_TRANSPARENT on flat texture flags when there are holes in the texture.
Minor fix to make sure MD2s always set the right blend mode
Pwease no kicky
Don't kick Tails! Also, a movement for the WRITESINT8 to prevent modification to buf if the function bails early.
This can go into Master, right? It only matters for the host, and it's explicitly only having a major effect outside of netgames.
See merge request !201
*Add CONS_Printf messages for !netgame checks
*Arg count is checked first regardless of netgame status for both kick and ban, < 2 is checked instead of == 1 just in case these weren't called from console for some stupid reason?
*Moved Command_Kick's buffer vars to within the code that actually does kicking stuff
Lua archive value fix
This fixes some potential problems with archiving mobjinfo_t/state_t data via Lua in netgames (whether as custom mobj/player vars or by NetVars hook). Thanks to LJSonic for pointing this out to me.
See merge request !200
Apart from the fact that UnArchiveValue reads UINT16 for both anyway (which alone causes problems), but UINT8 isn't even enough to store the higher end of the object types list and definitely most of the states welp
The issue was that because both them and the player had MF_SOLID, the tmfloorz of the spring was getting set to above the player (or vicea versa with tmceilingz), forcing it upwards with them under certain circumstances.
Now, springs only acknowledge the solidity (for purpose of tmfloorz/tmceilingz) of objects they CAN'T launch.
OpenGL slope FOF lighting fix
This fixes some issues with sloped FOFs that affect lighting in OpenGL (as in, those that cast a shadow or have a colormap). Particularly, they can do strange things to any wall textures adjacent to them, as we've noticed ourselves in levels for 2.2. =P
See merge request !194
Fixes with respect to sector special touching and slopes
Some important stuff.
* SF_TRIGGERSPECIAL_TOUCH now actually works. Previously, it abandoned the loop early if ANY bounding sector didn't have that sector flag, which it likely didn't - only checking one extra sector's worth of FOFs. Also, the teleport handling there is more robust, and actually bails out if you teleport, instead of just awkwardly continuing through the loop.
* SF_TRIGGERSPECIAL_TOUCH now works for each time thinkers, too.
* Fixed a bug with being able to go under lava because P_CheckSolidLava doesn't take slopes into account.
* Also, P_CanRunOnWater supports slopes now too.
* Quicksand supports slopes and reverse gravity now.
* Space Countdown supports slopes now.
Also, an experiment behind a #define which currently isn't turned on:
* UNDER A #define, "SECTORSPECIALSAFTERTHINK", WHICH IS CURRENTLY TURNED OFF, BUT I WILL WANT TO TURN ON IN INTERNAL: Moved sector touch handling to P_PlayerAfterThinker (from P_PlayerThinker before movement). Allows for being able to trigger moving slope sectors that are going down, most specifically lava (didn't matter in RVZS in 2.1 because you could clip through the sides and go underneath the lava, causing damage - a sloped testwad version of that prevented going underneath.) Also fixes one-frame standing on deathpits before you die. Basically means sector triggers effectively happen one tic earlier, since it's after movement.
See merge request !131
Polyobject seg render fix
This fixes both Software and OpenGL renderers so that polyobject segs aren't drawn if the game is drawing the actual subsectors they're from (outside the main level, where the polyobject walls were pre-spawn). They should only appear as part of the polyobject itself in-level.
This means a few glitches with polyobjects are probably fixed: for instance in Software mode, polyobject walls sometimes appear through level boundaries (and make everything above/below vanish, turning into HOM or skybox), if the BSP rendering code happens to find one of the subsectors said segs came from outside the level. I don't think anything similar happens in OpenGL, though I'm sure some unwanted typecasting is happening as a result of attempting to draw the segs. (And it fixes a crash in 2.2 anyway.)
See merge request !195
Ping-related code fix
Somehow, the part of the netcode for calculating the players' pings in a netgame frequently gets the concepts of nodes and players mixed up, which is probably not a good thing. This branch of course fixes those slipups.
I originally based this branch on master to be merged to it (since it only fixed issues on the host's side), but after finding another issue with clients receiving PT_PING from the server, I decided to make this a merge to next instead.
See merge request !193
Ld414 invalid sound fix
This fixes Linedef type 414 crashing the game if an invalid sound number was supplied to it (this can happen if you, say, scrambled THZ2's textures *cough*), whether or not the "Repeat Midtexture" flag is checked.
See merge request !196
Polyobject segs should ONLY be drawn if the polyobject itself is in the polylist of a subsector being rendered. That way you won't sometimes see polyobject walls through level boundaries, if you happen to be close enough to their pre-spawn locations outside the level (or in them, if you decided to go on a noclip journey).
Ignore askinfoviams
Turns out PT_ASKINFOVIAMS is an obsolete packet type anyway, the MS doesn't send it at all. So let's just ignore it in the netcode then. (the online page just acts like a client and sends PT_ASKINFO, and is not part of the MS itself)
See merge request !191
