Commit by entryway :: r4246 /branches/prboom-plus-24/prboom2/src/r_demo.c:

Check demo footer for errors and ignore it in case of corruption.

Commit by entryway :: r4245 /branches/prboom-plus-24/prboom2/src/gl_preprocess.c:

fixed compilation

Commit by entryway :: r4244 /branches/prboom-plus-24/prboom2/src/ (v_video.c v_video.h):

Submitted: J. Random Hacker ( jrhacker )

This is a clean-up of General fix for oddball aspect ratios - ID: 3522885, to be applied on top of that change. (1) Use a shorter and smarter implementation of ReduceFraction. (2) Note in comments that integer math is no longer used in FPS scaling so further simplifications might be possible.

Commit by entryway :: r4243 /branches/prboom-plus-24/prboom2/src/ (am_map.c gl_map.c gl_struct.h):

'nice map things' were not clipped by map_overlay_pos_(x, y, w, h) box

Commit by entryway :: r4242 /branches/prboom-plus-24/prboom2/src/am_map.c:

fixed compilation with msvc6

Commit by entryway :: r4241 /branches/prboom-plus-24/prboom2/src/ (7 files in 2 dirs):

'Max Health', 'Max Soulsphere' and 'Megasphere Health' DEH parameters did not work immediately after 'prboom-plus -warp x' since 2.4.8.1

Commit by entryway :: r4240 /branches/prboom-plus-24/prboom2/src/ (9 files):

General fix for oddball aspect ratios by J. Random Hacker (jrhacker)

I was on the road with an EeePC 1000, which has a native resolution of 1024x600, and PrBoom-Plus did not get the aspect ratio right. PrBoom-Plus 2.5.1.3 uses a static table of common aspect ratios that includes 16:9 and 16:10, but 1024x600 is 128:75 (not quite 16:9). Rather than just add another line for 128:75, I replaced the static table with a function that computes the needed parameters dynamically based on arbitrary window geometry and user preferences. The VGA modes 320x200 and 640x400 are recognized as special cases, while all other geometries work on the assumption of square pixels.

Commit by rjy :: r4239 /trunk/prboom2/src/lprintf.c:

Attempt to compensate for lack of va_copy

Pretty much copied out of git, see explanations/patches here:

https://git.kernel.org/?p=git/git.git;a=commitdiff;h=ab8632ae
https://git.kernel.org/?p=git/git.git;a=commitdiff;h=26db0f2e

Thanks also to entryway for the help.

Commit by rjy :: r4238 /branches/prboom-plus-24/prboom2/src/ (e6y_launcher.c e6y_launcher.h):

chmod -x e6y_launcher.{c,h}

Commit by rjy :: r4237 /trunk/prboom2/src/d_main.c:

Fix const mismatch warning introduced in r4235

Commit by entryway :: r4236 /branches/prboom-plus-24/prboom2/src/p_enemy.c:

forgot to remove test code

Commit by rjy :: r4235 /trunk/prboom2/src/ (9 files):

Avoid use of PATH_MAX

Get storage from malloc instead, calling snprintf twice to get the sizes
needed in most cases.

Also, bug fix: in FindResponseFile, when looking in I_DoomExeDir for the
file, it was not correctly path-separated from the directory. Previous
code merely appended it, which is not correct, as I_DoomExeDir's return
value does not end in a slash. That this went undiscovered for so long
shows how much response files are actually used, I think. Note you can
achieve the same effect of a parameter @rsp with $(<rsp) in the shell.

Commit by rjy :: r4234 /trunk/prboom2/src/ (6 files in 2 dirs):

doom_snprintf: attempt to work around non-standard stdio

Based on information gleaned from searching MSDN for snprintf,
it seems that windows' implementation:

• on buffer overflow, returns -1 (instead of the number of characters that would have been written if the buffer had been big enough.)
• on terminator-only overflow (i.e. the only character there was no room for was the terminator itself) it does return the number of characters written, but fails to null-terminate the buffer. This is an attempt to write a wrapper that works around these issues.

Commit by rjy :: r4233 /trunk/prboom2/ (9 files in 6 dirs):

Assume snprintf exists everywhere

• All platforms' config.h files (VisualC6, VisualC8, MAC) say it exists (on the former two it has a different name, which is #defined)
• POSIX/i_system.c already assumes it exists, with no #ifdefs
• Chocolate Doom assumes it exists, with a similar windows #define
• Even other highly portable C projects such as Git assume it exists (although Git doesn't assume it has a sane return value)
• Removing the ancient workaround, calling sprintf when snprintf is unavailable, can only be a good idea! Better to have bug reports about compile failures than security holes on people's systems...

Commit by rjy :: r4232 /trunk/prboom2/src/SDL/i_video.c:

Ignore spurious mouse motion events after grab state change

Move out code that handles mouse grabs into its own function,
make that also handle flushing the mouse motion event queue,
and call it from both I_StartTic and I_UpdateVideoMode.

Commit by rjy :: r4231 /trunk/prboom2/src/ (SDL/i_video.c d_main.c):

Ignore spurious mouse motion events after screen mode change

• I_UpdateVideoMode: call SDL_PumpEvents/SDL_PeepEvents after a mode change to remove spurious mouse events from the queue
• D_PostEvent: remove the kludge ignoring events for small gametics

Commit by entryway :: r4230 /branches/prboom-plus-24/prboom2/src/ (7 files):

reworked player->centermessage code

Commit by entryway :: r4229 /branches/prboom-plus-24/prboom2/data/ (13 files in 2 dirs):

Added PLS1 and PLS2 sprites from MBF "-beta"

Commit by entryway :: r4228 /branches/prboom-plus-24/prboom2/src/ (7 files):

Added "-beta" stuff from MBF (A_FireOldBFG, A_BetaSkullAttack, A_Stop)

Commit by entryway :: r4227 /branches/prboom-plus-24/prboom2/ (11 files in 3 dirs):

stuff