[AD] Allegro 3.9.31

[Shawn Hargreaves <shawn@xxxxxxxxxx>]

... is on http://www.sunsite.auc.dk/allegro/wip.html

My apologies for taking so long to get around to this: I've been 
considerably busy with other stuff recently.

I also haven't done very much work other than merging other people's 
patches, and a few bugfixes. In particular the nasty problem with Windows 
screen sub-bitmaps is still outstanding, but don't worry: I will get 
around to that eventually, I promise :-)

On the plus side, this is the first ever version of Allegro to be 
officially certified as Y2K compliant! (unlike many vendors, I'm 
pessimistic enough not to have claimed that until I actually saw it 
working in ALLEGRO_DATE_STR, the grabber update function, etc :-)

Michael: I noticed that with your X patch to change the keyboard repeat 
behaviour, this seems to be global for all X apps, which is quite 
disconcerting! Perhaps we need to switch back into normal repeat mode 
whenever the Allegro window loses input focus? Strange that X doesn't do 
this automatically: maybe it's just something odd about my setup?



Changes since 3.9.30
--------------------

   Greg Hackmann contributed a Borland C++ Builder port. Like the RSXNT 
   and Mingw32 versions, this is a client for the MSVC DLL, and not able 
   to compile Allegro directly itself, but allows you to build Allegro 
   programs using the Borland compiler.

   Henrik Stokseth did some work on a native Mingw32 build. This is 
   currently incomplete, and unlikely to work right now since there have 
   been other changes in the Windows code since the version he started 
   from. At the moment this version is statically linked, using 
   makefile.mgs, and makefile.mgw still exists for the MSVC-piggyback 
   version. The ultimate goal is to get this native version building as a 
   DLL in addition to statically linked, at which point we can merge both 
   makefiles into makefile.mgw and remove makefile.mgs. Oh yeah, and 
   there are currently no docs for the native version.

   Isaac Cruz added support for windowed modes in color depths other than 
   the current Windows desktop, and hardware accelerated system bitmaps.

   David Kuhling optimised the i386/gcc fsqrt() code, and added an 
   fhypot() routine.

   James Hyman added support for dithering truecolor down to paletted 
   images.

   Robert Ohannessian added MMX optimisation for the 8 and 16 bit clear() 
   functions, and Jose Antonio Luque optimised the 16 bit blit() and 
   masked_blit() routines.

   Vincent Penquerc'h added a yield_timeslice() function (currently 
   implemented in DOS and Unix).

   Added a desktop_color_depth() function (currently only implemented in 
   X: a Windows version will follow).

   Jose Antonio Luque improved the Windows joystick driver.

   Chris La Mantia added textout_right(), textprintf_right(), and 
   textprintf_justify() functions.

   Sven Sandberg added a ustrrchr() function, and improved the usprintf() 
   handling of large floating point values.

   Peter Pavlovic added a Slovak keyboard mapping and message translation.

   Manni Heumann fixed some problems with the German keyboard mapping.

   Michael Bukin improved the X keyboard and mouse handling, and fixed 
   some problems with building on non-GNU Unix systems.

   Peter Wang added input support to the Linux OSS sound driver.

   Many bugfixes and tweaks from various people (sorry, my typing 
   patience has run out so you don't get a full list :-)



Todo before 4.0
---------------

   The DirectDraw windowed mode driver seems to get confused about 15 vs. 
   16 bit modes in cases where Windows reports a 5.5.5 mode as being 16 
   bit (which is not how Allegro describes it). Need to investigate this. 
   Also windowed modes with mixed color depths just show up black on my 
   system: need to investigate that, too.

   Implement the desktop_color_depth() function in Windows.

   Implement scheduler timeslice yielding in Windows.

   Nasty windows bug involving screen sub-bitmaps, some sort of locking 
   problem methinks. Related to this is that graphics.c should clear the 
   BMP_ID_LOCKED flag during sub-bitmap creation (but doing this makes 
   the aforementioned bug worse, which is why I left it out of this WIP).

   Someone (me? urgh, starting to look that way...) needs to install 
   Mingw32 and get the native build working 100% both static and as a 
   DLL. Also need to rationalize the code to remove duplicates of MSVC 
   and gcc format inline asm.

   It would be nice if Windows sample looping worked properly.

   Add caching system to speed up filename#object and 
   load_datafile_object() (damn, this one has been in the list for nearly 
   a year now :-)

   Talk to Calin about merging latest P3D rendering stuff (and this one 
   too: I haven't heard anything from Calin for so long, maybe it's time 
   to drop this todo entry entirely...)




-- 
Shawn Hargreaves - shawn@xxxxxxxxxx - http://www.talula.demon.co.uk/
"A binary is barely software: it's more like hardware on a floppy disk."