Build 5375
SVN-
Generale
Fix a typo in loadpalette that broke reading of the lognumalphatabs byte.
Nickname SVN e nomi Git sono unificati per persona.
Fix a typo in loadpalette that broke reading of the lognumalphatabs byte.
Fix error in r5346 that broke the MoreBlendTab section of PALETTE.DAT.
Fix RENDERTYPE=WIN build by changing a const char* to a char*, which had been cast to (void *) before passing to free() (which is bad practice) prior to r5352.
Build tools/OS X: Fix build by satisfying compat.c's dependence on osxbits.m.
Build tools/Windows: Fix text outout by compiling without -mwindows.
Fix a bug with the tilefromtexture { texture } subtoken, where alphacut was interpreted in an inverse fashion, generally causing hightile textures defined this way to appear completely transparent.
Correct handling of size-zero gamearrays: Prevent a crash upon loading a savegame containing them, and allow resizearray to zero.
Fix a harmless "Internal compiler error" warning when using quick structure access on userdef.
after playtesting with Esc+[T], break from the key polling loop.
Bug noticed and fix located by The Mechanic.
Fix GTK build.
Fix USE_OPENGL=0 build.
Mostly revert r5326, and instead restrict the starting and stopping text input to touch devices where we need the on-screen keyboard to appear. This maintains the bugfix on targets with a physical keyboard.
Fix the Custom game content directory startup window option loading the folder alphabetically previous to the one selected when the "autoload" folder is present.
fix oob write for Ctrl+K ("delete all to end of line").
add config variable 'corruptcheck_heinum'.
With 3 modes: 0: auto-correction and warning disabled 1: auto-correction enabled (default, current behavior) 2: auto-correction and warning enabled
Revert r5316 which completely broke BOUNCEMINE.
don't spawn FRAMEEFFECT1 for BOUNCEMINE.
There's no point, really. Since the bouncemine is immobile, the tsprites will all be overlaid at one point and show the same picture, as far as I can see. Worse, since r4876, the generated sprites may draw on top of the actual one. The former have their shade always set to sprite[t->owner].shade though, and will then appear fully bright (shade -127, since BOUNCEMINE sets it like this). This fixes issue number 4 from http://forums.duke4.net/topic/8042-five-minor-problems-report/
fix displayed viewscreen when its x/yrepeat isn't evenly divisible by 4.
fix unbalanced begindrawing()/enddrawing() pair in printext16().
Fix various, but not all warnings.
Except: - missing braces around initializer [-Wmissing-braces]: due to union-in-struct vec3f_t - comparison between signed and unsigned integer expressions [-Wsign-compare]: in polymer.c
fix frame locking issue with 2d3d mode, crash when toggling bottom texture swap, issues with drawing certain types of walls.
fix display of pal 4 sprites in 2d mode.
fix 3d mode status messages being overwritten with position/ang/horiz readouts.
fix unreadable status messages in 3d mode when working with a palette that doesn't align to Duke3d's color layout.
don't warn about the map being corrupt if the only thing wrong with it is a sector ceilingheinum or floorheinum set without the corresponding bits to enable sloping set. DONT_BUILD.
Fix warnings
fix regression with Y-flipping bit.
fix synthesis warning.
fix two "variable may be used uninitialized" warnings.
I didn't produce a case when this could actually have happened, but reading the code, in the case of variable 'j' it seems possible with an empty map.
fix a possible oob read via drawspritelabel; sanitize 2d3d mode.
That is, in m32_is2d3dmode(), also check if 2d and 3d mode resultions are the same. Otherwise, oob writes (e.g. via tileinfo_doprint()) and other thinkable niceties may ensue.
fix 3d picture-in-picture when editing a map with TROR.
Address complaints from the "Mapster32 problems and bugs" thread.