/* * CHANGELOG. * *  Please do NOT harrass Ken Silverman about any code modifications *   (including this file) to BUILD. */07112000 - Initial release from icculus.  C code compiles, doesn't link, lots            of bastard assembly.07122000 - LOTS of ASM cleanups by Andrew Henderson.07132000 - Ryan adds some missing functionality: SDL equivalents for most of            the VESA functions, filelength(), and some other piddly stuff.07142000 - Dan Olson cleans up the code a bunch, fills in a bunch of the            ASM routines with C equivalents, and puts some VESA replacements            in place. Other good stuff. Ryan cleans up some, too, put some            more stubs/implementations in place, and continues to flesh out            the SDL driver.07212000 - A BUTTLOAD of updates. "build" links, and segfaults, because I            screwed up all the inline ASM that I did. Will be correcting,            shortly. "game" doesn't link yet. Haven't looked at it in detail.            Moved anything with i/o port calls into dos_driver.c, and put            stub equivalents in sdl_driver.c ... The setupmouse() function got            commented out because it was getting called before the video got            setup, I think. Will check.  (never released.)07272000 - Assembly in pragmas.c was borked (thanks to RYAN.), and is now            halfway fixed. Ryan will be fixing this completely Real Soon Now.            Other ASM may be screwey, too, if Ryan touched it, but the NASM            code that Andrew did is probably fine (mostly because he knows            what he's doing... :)  )07282000 - Fixed borked ASM code. Fixed a text file reading routine that            depended on finding DOS-style newlines. Moved some more stuff to            sdl_driver.c and dos_driver.c ...07292000 - Fixes in pragmas.c by Dan Olson, who also knows more about this            inline ASM stuff than Ryan. Dan also cleaned up engine.c's ASM.            Thanks to Dan and Ryan, game.c now compiles, albeit with a            thousand warnings. "game" does not link yet.07312000 - Thanks to Dan, "game" now links, other minor fixups by Ryan. Broke            up CHANGELOG into some standard docs. (README, CREDITS, etc...)08012000 - Andrew does some graphic porting. Looks like we've got blocks where            fonts should be, and a mouse cursor. Huh. Ryan abstracts 0xa0000            addresses to VIDEOBASE define. Under non-DOS platforms, this is            currently "surface->pixels". Basic events were hooked up. Added            FreeVGA to README. Lots of fun new segfaults.08042000 - Ryan gets the keyboard working apparently fully, and the mouse            sorta. The grid now draws correctly. No more segfaults on exit.            You can zoom in and out on the grid with the 'Z' and 'A' keys.            Fun, huh? sdl_driver's drawing routines (most notably, drawline16)            are MUCH more robust. Some minor structure changes. Other things            seem to work within build (space draws a line from a point to the            mouse, etc.), but I dunno what any of this means; must find a            BUILD HOWTO. Probably other stuff. Timer is hooked up, but SDL            won't fire a timer faster than every 10ms, or 100 times a second.            BUILD expects this timer to fire every 8.3~ seconds, or 120 times            a second. This may be problematic in the game, but it probably won't            be a problem in the BUILD editor, where it just handles keyboard            updates. An alternate timer system might get swapped in later.            Game still links, but aborts on initing the timer, since I need to            sync with the editor; Ken saw it fit to commit the mortal sin of            cut-and-paste instead of having a unified BUILD_Init() function,            so I need to update it to the new structure (using sdl_driver or            dos_driver, calling _platform_init(), blah blah blah...)08052000 - Dan fixed an ASM screwup in getkensmessagecrc(). Ryan fixed the            mouse button event code. Added event thread, and some bits of            thread safety in sdl_driver.c. Mouse now seems to move okay if the            X11 cursor is hidden, and we grab the input. Goes to fullscreen            and grabs input by default; export BUILD_WINDOWED=y and            BUILD_NO_MOUSE_GRAB=y, respectively, to disable. Got myself a            copy of Duke3D Atomic Edition for the data files. Fixed the            #pragma packs in build.h. Fixed _platform_init() functions.            Improved SDL driver. Commented out a bunch of self-modifying ASM.            Dan got the fonts to display. Ryan got the correct palette loaded.            Lots of other little cleanups. Good stuff.08062000 - Ryan fixed clear2dscreen() in SDL driver. It's (hopefully) correct,            now. Also, all that missing text should (hopefully) now display.            drawline16 is more correct now. Other little cleanups and            move-arounds. It seems that "pageoffset" is the big enemy around            here today. Hum.08072000 - Ryan gets the BUILD title text to stop flickering. Prints our names            at exit (maybe someone will want to hump OUR legs, too). Added the            _idle() function to dos_driver and sdl_driver, and sprinkled some            calls through overheadeditor() in build.c ... ideally, we'd            eventually want to not have the editor wipe the surface and            redraw everything like mad constantly, but rather wipe and draw            as needed to choke the processor even less. Added implementation            of getfilenames() in build.c...seems to work well enough. Got            map loading dialog to appear (in the wrong place), but the arrow            keys don't seem to work to test further. Added caption to window's            title bar. Dan completely the Herculean effort of cleaning up            EVERY compiler warning in the editor. Other tinkering.            Adrian Neill and Christian Zander cleaned up all the compile            warnings in game.c and multi.c. Christian also supplied a shell            script to set the BUILD_* variables for developing. Ryan fixed            permissions of map files in saveboard() in engine.c. game.c had a            static variable called "sync" that conflicted with unistd.h's            sync() function, so the static's been changed to _sync. Fixed            more graphic stuff; pageoffset is just blindly set to zero on            Unix, now, and that seems to fix some stuff. printext16() was            changed to statusbar_printext16() in the appropriate places, so            I could remove my ypos += 336 hack in the actual implementation,            make this #define to regular printext16 on DOS, and be a one-line            function to add 336 to ypos in Unix. sdl_key_filter now has            support for extended keys; in init_new_res_vars(), just add keys            with values of 0xE0** where "**" is the normal value. This allowed            me to add the arrow keys and the keypad enter key. Tried loading            Ken's maps, and they WORK. With today's graphic fixes and the            arrow keys, you can select a file from the loading menu, and            navigate around in 2D mode. Cool. Switching to 3D mode segfaults            immediately.  :)  Changed sdl_driver to NOT double buffer, and            the SDL_Flip() call is now a screen-wide SDL_UpdateRect(). Added            Ken's maps for his test game to the archive.08082000 - Added key bindings to sdl_driver: ALT-Enter (or ALT-Return) to            toggle between windowed and fullscreen, and CTRL-G to grab/ungrab            the mouse pointer. Increased sensitivity of mouse movement. Dan            fixed the editor's status bar; it now draws. I think Dan also            squashed the rest of the pageoffset annoyances in the editor.            Dan tracked down the line drawing problem, and Ryan patched some            ASM (clearbufbyte(), specifically), so you can add walls, now.            Things are REALLY starting to look good.08092000 - ASM day. Fixes and fights from Dan and Ryan, with lots of wisdom            from Andrew. Realized that we were in violation of Ken Silverman's            license by not updating/including his copyright header on all            source files. This has been fixed; sorry, Ken. Other video stuff.            Tweakage. 3D mode no longer segfaults, but is still garbage.08112000 - Fixed a bug; ALT-Enter wasn't changing sdl_flags, so next time you            changed resolutions, you went back to your original (non)windowed            state. Did the same for mouse grab state. Dan converted the ASM            for some more engine.c functions, and added a missing key to our            SDL scancode table. Hooked up keybindings in the (still not            functional) 3D mode. Discovered that we were missing some data            files. More sdl_driver tinkering. Moved stuff out of dos_driver's            _nextpage(), and back into the abstract engine.c section.            I don't know what I changed, but the sprite and vector squares            don't draw in the correct place anymore in engine.c's            draw2dscreen() ... culprit is something in sdl_driver.c...someone            diff this version against the last. Grrr. Generally, I'd fix this            before sending out a tarball, but it's 5:00a.m., and I need to get            something out the door here. Will fix for tommorow, and it doesn't            otherwise impede your use.08122000 - Turned optimizations back on, and the broken drawing from last            night works again. Yikes. Dan stubbed out lots of the missing C            functions in a.c (someday, we'll fill really these in, I hope.),            and made mad progress on attacking our palette issues. Some compile            fixes (incomplete) to allow building without SUPERBUILD defined            (for removing voxel support...might be needed for pre-voxel games            like Duke3D...) The self-modifying ASM passes function parameters            in registers, so I needed to do an inline wrapper. This also now            sits in a.c.  (Still crashes and otherwise doesn't WORK the same            in 3D mode, but still...) Added tiles000.art to the archive from            Duke3D shareware's duke3d.GRP file. I hope this doesn't violate            anyone's copyrights.08132000 - (For reference, apparently GCC requires -O2 for inline asm to work.            That would explain the drawing barfs...)  Dan fixed the status bar            (after Ryan broke it again.  :)  ), and moved krecipasm() into            a_linux.asm, since FPU opcodes don't inline correctly. Added Ken's            stuff.dat. As a milestone, 3D mode now seems to work a LITTLE.            Looking in the wrong place causes a segfault, and generally, what            very little draws seems mostly incorrect, but progress is being            made. Changed out all the collected data files with Ken's original            STUFF.DAT. initgroupfile() now aborts the program if it can't            open a data file. (Without stuff.dat, you get a blank surface, and            you have to blindly feel your way out of the program.) Dan also            started work on the command line tools, like kextract. Those are            in a separate tarball on the website.08152000 - (Stuck at LinuxWorld...sorry for delays.) 3D mode now does not (by            default) segfault, and you can stumble around. Some walls seem to            render correctly. Lots of things cause segfaults. Disabled the MMX            overlays for now, as that is a little less self-modified code in            use. Dan fixed some keyboard scancodes (and also independently            fixed the same 3D mode things. hah. :)  )08192000 - Back from LinuxWorld. Added an obnoxious hall of mirrors detector            for 3D mode. _asm_slopevlin now returns immediately, which for now            stops most 3D mode segfaults. This will have to change, but it            helps for now. Put in a hack for checking frames-per-second.            Added some new keyboard scancodes.08212000 - Another keyboard scancode added by Dan. Changed 3D mode to prompt            to save changes before dropping out of graphics mode; the stdio            thing was pretty obnoxious. Now we are basically waiting on a copy            of Watcom C to show up, so I can trace through the original code            and verify that I'm getting sane parameters and return values.            Needless to say, productivity has dropped off quite a bit.08232000 - Dan found a screwup in the inlined dmulscale32()'s return value.            Consequently, now the hall of mirrors are gone. Cool. Walls seem            to render correctly all the time now. Other visual problems, and            no floors or ceilings, but this is a BIG win. Thanks, Dan! (A few            other ASM functions in pragmas.c had this problem. dmulscale32()            was the most noticable fix, though.)08242000 - Dan added/fixed some more ASM. The SDL parachute is now in place,