* Replace the texheight parameter provided directly to it with a previously existing (now renamed) global used for the same purpose, so that it can be used as an interchangeable function pointer with R_DrawMaskedColumn.
* Using the above, optimise R_DrawVisSprite to call a function pointer in a tighter loop rather than check SC_VFLIP each time around.
* SHORT macro can involve repeated operations; calculate once and put in memory for both RANGECHECK and papersprite.
* Remove irrelevant range check (already covered by existing range check immediately above) from R_DrawFlippedMaskedColumn and R_DrawMaskedColumn.
* "Warning: avoiding a crash in %s %d" is a terrible error message, and it chips away at the tightness of the loop just for something most people will never see printed. Replace with a PARANOIA I_Error in case someone actively wants to go hunting for its cause.
I heard properpaper had some weird crashes? I couldn't reproduce
them no matter how hard I tried, but I added some bounds checking
to this version too just in case. Gotta get other people's help
to try to reproduce those.
Also:
* SPR2_XTRA - instead of defining lumpnames in S_SKIN, those kinds of assets can just be bundled into the spriteset. Required for ending cutscene stuff, I guess, but also done for HUD life icon and character select image (aside from Sonic&Tails, still SOC'd in).
* Minor oversights in SPR2 support corrected.
* Better evaluation, featuring ending assets.
* Intro has warping-in blackrock, reusing ending assets.
* Cutscene text now supports lowercase (intro and custom).
* Disable the asset-fucking "gamma correction" I put in over two years ago when implementing colour cube. (This is the only thing I could move into another branch if you MUST, but it's basically invisble in the diff so w/e.)
* Don't blank the screen if the top left pixel of a screen-covering patch is transparent. (Checked via nonzero topdelta for first column)
Bugs:
* OPENGL ONLY: The first ~20 frames of both endings are fucked. A little help here? Might be HWR_DrawFadeFill's fault, which I just created. OR it could be in f_finale, but I doubt it, since it doesn't appear in Software.
Vissprites are now only clipped against their respective portal's geometry obtained from their BSP run.
Additionally, if a portal is provided, they're clipped to the portal's clip boundaries.
The work on this branch should conclude after a pair of remaining glitches are fixed.
The drawnodes are now fully grouped in separate lists, and then sorted individually. This fixes sorting problems caused by portals belonging to differently perceived scales (skyboxes for example).
Drawsegs and vissprite/drawnode sorting require the viewz, so the viewz is stored for each portal/view, and then restored when needed; without this, the rendering process erroneously sorts the elements, and draws some at wrong positions.
Each shall eventually have its specific vissprites/drawsegs; currently only drawsegs are stored in their correct list, vissprites are stored in the first list as a placeholder.
The idea is to sort each list individually, and then render their masked elements, starting from the last drawnode list.
This retains a non-recursive function calling method while still rendering things in order.
