The ovl file is platform independent (it runs inside dosbox, so it can be on Z: ), but libglide2x should be a wrapper (unless you have a real voodoo card...although I only tested this with windows98 😀). I'm not sure how many linux wrappers exist, but at least one does and has been tested to work: OpenGlide. Be sure to build SDL version though (you might need to fiddle with the Makefiles), also, search for the functions with the __fastcall calling convention and remove it (the __fastcall keyword, function stays 😀)

Hi gulikoza,

thanks a lot for your quick answer! I'll try following your suggestion, and I'll keep you updated on the results in a later reply.

Wow works like a charm, thank you!

It took me a while to understand that I needed the CVS version of openglide (the zips on the site are old and no Makefile in them), but then I managed to compile the SDL version as you suggested.

The 3dfxdiagnostic tests work fine,but they freeze dosbox when pressing 'A' to leave them: Tomb Raider, instead, works very well and quits fine.

Now I only have to get back the Redguard CDs I lent to a friend of mine and cross my fingers 😀

Thanks again, cheers!

Sorry, I forgot to mention cvs is needed 😀
I have actually started to prepare some patches for openglide to improve the build process (and include some performance fixes), but haven't had the time to finish them yet...
RedGuard should work fine, but if it doesn't start, try running it with a low cycle count (say 10.000-15.000 cycles) and increase the cycles when the loading screen appears...

I read about the low cycles when loading in some old discussions about Redguard. However, I didn't need to do it, I left cycles=auto and just had to set the memsize to a high value (I've chosen 64 MB) and all went fine.

All the intro videos play well, while the game, even if good looking, is really slow (kind of 0.5 fps, I guess). I don't know if this is normal: the linux build is almost an year older than the windows one, so probably much has changed in the meanwhile...

Well...those openglide performance fixes I was talking about help a lot 😉

I just tried the patch, and there are two results:

1) now I can see the static background image while loading the scene after the videos, which was missing before;
2) but after the loading process, I just see a pink block! nothing else... even if I can hear the game running in the background (specifically the two pirates beating me 😀 )

Hmm..I was pretty sure I've nailed all the bugs. Perhaps this one is linux specific (hopefully not nvidia specific, since I hava a radeon 😜)

Let's hope you're right, considering I have an nvidia 🙄
Let me know if there's any way I can be of help in debugging this problem...

Well...Windows version seems to work and my linux is a mess 😜 I can't really get opengl to work...but...
I have just released the sources of my patch. Try compiling your own dosbox version (the ykhwong's linux version is quite old...it is possible that the ovl file and dosbox got out of sync 😀)

Hi,

sorry for my late answer, I missed your reply... thanks for releasing your sources!

I just finished compiling your latest CVS, dated 23-06-07. In order to be of help to others who might want to accomplish this under linux, I had to make the following things before:

• 1) I removed any reference to hq2x_d3d, since there's no Direct3D in linux;
  
  2) I renamed ScalingEffect.* to scalingeffect.*, because otherwise the compilation would complain (in linux files are case-sensitive);
  
  3) I had to make use of gcc/g++3.4, because 4.1 was complaining about some asm commands.

However, how do I enable glide emulation? Is it enabled by default or is there some line I have to add to dosbox.conf as it works in ykhwong's build?
I just tried the 3dfx diagnostic tests and they do nothing: I start the test by pressing the 'A' key, and it does nothing; then when I press a key (that is how you exit the test when it's running) the console becomes active again. I had this behaviour when testing your windows executable with Wine as well.

Hope to hear you soon, thanks in advance!

Sorry, I missed your other post with the glide patch... ignore this message, I'll try that one 😀

Hi,

Thanks for reporting, I'll fix those in my patches, so that my sources will compile cleanly. Where was gcc complaining about asm? In MT32?

Yes, this is the exact error it gives me when compiling with 4.1:

if g++ -DHAVE_CONFIG_H -I. -I. -I../.. -I../../include -I/usr/include/SDL -D_GNU_SOURCE=1 -D_REENTRANT -g -O2 -MT file.o -MD […]
Show full quote

if g++ -DHAVE_CONFIG_H -I. -I. -I../.. -I../../include -I/usr/include/SDL -D_GNU_SOURCE=1 -D_REENTRANT -g -O2 -MT file.o -MD -MP -MF ".deps/file.Tpo" -c -o file.o file.cpp; \
then mv -f ".deps/file.Tpo" ".deps/file.Po"; else rm -f ".deps/file.Tpo"; exit 1; fi
../../include/structures.h:154: warning: ‘packed’ attribute ignored for field of type ‘MT32Emu::TimbreParam [8]’
../../include/structures.h:156: warning: ‘packed’ attribute ignored for field of type ‘MT32Emu::PatchParam [128]’
if g++ -DHAVE_CONFIG_H -I. -I. -I../.. -I../../include -I/usr/include/SDL -D_GNU_SOURCE=1 -D_REENTRANT -g -O2 -MT i386.o -MD -MP -MF ".deps/i386.Tpo" -c -o i386.o i386.cpp; \
then mv -f ".deps/i386.Tpo" ".deps/i386.Po"; else rm -f ".deps/i386.Tpo"; exit 1; fi
../../include/structures.h:154: warning: ‘packed’ attribute ignored for field of type ‘MT32Emu::TimbreParam [8]’
../../include/structures.h:156: warning: ‘packed’ attribute ignored for field of type ‘MT32Emu::PatchParam [128]’
i386.cpp: In function ‘float MT32Emu::iir_filter_sse(float, float*, float*)’:
i386.cpp:142: error: unknown register name ‘xmm3’ in ‘asm’
i386.cpp:142: error: unknown register name ‘xmm2’ in ‘asm’
i386.cpp:142: error: unknown register name ‘xmm1’ in ‘asm’
make[3]: *** [i386.o] Error 1

The error disappears if I make use of 3.4 instead.

Besides, there are another couple of issues I forgot to report in the previous mail:

1) the attached configure script doesn't have executable permissions, it must be made executable with a 'chmod +x' in order to work;

2) there's a typeless size variable in drive_physfs.cpp:

if g++ -DHAVE_CONFIG_H -I. -I. -I../.. -I../../include -I/usr/include/SDL -D_GNU_SOURCE=1 -D_REENTRANT -g -O2 -MT drive_physf […]
Show full quote

if g++ -DHAVE_CONFIG_H -I. -I. -I../.. -I../../include -I/usr/include/SDL -D_GNU_SOURCE=1 -D_REENTRANT -g -O2 -MT drive_physfs.o -MD -MP -MF ".deps/drive_physfs.Tpo" -c -o drive_physfs.o drive_physfs.cpp; \
then mv -f ".deps/drive_physfs.Tpo" ".deps/drive_physfs.Po"; else rm -f ".deps/drive_physfs.Tpo"; exit 1; fi
drive_physfs.cpp: In constructor ‘physfsFile::physfsFile(const char*, PHYSFS_file*, Bit16u, const char*, bool)’:
drive_physfs.cpp:648: error: ‘size’ was not declared in this scope

Changing line 648 to

Bit32u size=(Bit32u)PHYSFS_fileLength(fhandle);

fixes the problem.

With these changes, it seems to work fine! Of course I didn't test it thoroughfully...

Zip does not store permissions 😀

I'll leave the mt32 error, I have no idea how to fix it...the other erros have been fixed.

The MT32 error is due to a misdetection of SSE support. the code uses SSE assembler, but the compiler flags do not enable them. -march=pentium3, -msse or similar should do the trick, although the real bug is probably due to a wrong #define clause that doesn't recognize SSE is unsupported in that compile.