What changed in EDuke32, build by build

Clear GL depth buffer before drawing a frame in Polymost. This should fix the issue with skies in the Duke64 TC.

Revert "Polymost "use tile shades": don't confound shade and visibility." Temporarily add a r_usetileshades 2 to switch between Helixhorned and myself's ideas of how this should work.

Fix a weird problem where the arrow keys would sometimes send a shift keypress on Windows, manifesting in odd things like inconsistent turning speeds when using the keyboard, the grid helper lines showing up in Mapster32 when they weren't supposed to, etc. For some reason, an extra raw input keyboard event is generated with an out of bounds VKey value (0xFF) which must be filtered for proper operation. I wasn't able to find any official documentation regarding this but I did find it mentioned in a couple of random newsgroup/forum postings.

Unify glow and detail texture setup between polymost and mdsprite

Make the menus look the same in all rendering modes, because the LOADSCREEN tile looks like shit when any kind of palette-based shading is involved. So, now everyone gets the ugly metal looking MENUSCREEN.

Get rid of that old model occlusion checking hack I wrote for Polymost years ago. I actually didn't know this was still in EDuke32... it's one of the worst things I've ever written. :p

A bit more work on the Polymost shade table mode (r_usetileshades)

Add an experimental shading mode to Polymost for .art tiles, in which extra textures are allocated for each shade level based on the classic mode shade tables. Enable with r_usetileshades 1 (disabled by default). This uses a lot of memory and is still very WIP (a lot of the shading is completely wrong at this point).

Fix issue with Polymer checkbox in startup window, as reported at http://forums.duke4.net/topic/2961-paper-cuts-minor-bugs-and-annoyances/page__view__findpost__p__158208

Automatically execute mapname.cfg at map load time. This allows different maps in mods to have different rendering settings (shadescale, etc). This is not currently suitable for standalone user maps as there's no functionality to reset such cvars back to default at the next map load (so, if using this in a mod, every map needs to have a cfg).

Mask the seam that shows up when drawing the scuba HUD overlay in OpenGL by drawing SCUBAMASK 3 times... once at the left position + 1, and then at the proper left and right positions. It's definitely a hack, but at least it lives on the game side of things.

Fix changing the max refresh rate in EDuke32 (it was already working correctly in Mapster32).

Limit attempting to toggle DWM off when using OpenGL modes to Vista/7, because Windows 8 no longer supports disabling DWM. Apparently, Windows 8 contains a software DWM renderer, and all of the "Metro" bs is directly tied to DWM, so there is no longer a way for an app to disable it at will.

Add "ulotag" and "uhitag" to get/setactor/wall/sector, which exposes the tags unsigned, without the typecast to int16_t. This might be useful to someone.

Fix the warnings when building with C++, add MSVC C++ build support. This also changes the internal type of lotags, hitags and cstat type fields to uint16_t instead of int16_t to clean up some dubious behavior wherein the code was using a value of 32768 as a flag in these fields for certain types of things, like invisible sprites (using the value as if it was uint16_t) where it was elsewhere checking if the value was < 0 (using the value as if it was int16_t). This change may break a few specific effects if any part of the relevant code was missed when looking for areas that needed to be addressed.

Comment out the A_GetZLimits() call right before the Bassert() in A_MoveSprite(). This call is a problem because the assertion expects "dasectnum" and spr->sectnum to match, but A_GetZLimits() can itself call A_SetSprite() which results in another call to A_MoveSprite() and can potentially change the sectnum again.

Fix a.masm changes so MSVC builds work again. The relevant functionality seems to work (non-power of 2 textures drawn without error), even with CLASSIC_NONPOW2_YSIZE_WALLS defined in engine.c, but I'm not familiar with any corner or edge cases that need to be watched out for with this.

"Band-aid" fix for the issue with EDuke32 and Mapster32 visually locking up but continuing to run and play sounds after alt+tabbing away for a few seconds on Windows Vista/7. This may screw something else up, so please test thoroughly.

Addition of 3 events:

Add a couple of missed sanity checks for accessing player and sprite structure members as vars. Should fix crashes with stuff like "ifvare player[some value greater than MAXPLAYERS].kickback_pic blah"

Cleaning out my tree... mostly internal changes. Adds workaround to disable texture compression with the crappy fglrx driver on Linux, fixes FIRE sprites so that they don't render at their sector's floorz all the time, changes CON compiler around a bit to use a loop instead of calling C_ParseCommand() 10 million times.

Disable texture compression when ATI's horrible OpenGL implementation is detected

Add the svn revision to the version string in Mapster32

Consolidate POLYMOST preprocessor define into USE_OPENGL, remove SUPERBUILD preprocessor define, add additional mode to neartag() to skip sprite searches and speed up processing (and enable for CON_OPERATE), fix issue with nearby single instance sounds not playing due to out of range sounds blocking their playback, fix issue with settings.cfg being reset to default bindings when running a mod that specifies its own cfg base name, improve CON structure member interface read/write performance, fix a bug with CON "move" pointer validation