.mdl loading (CAnimatedMeshHalfLife.cpp) on big endian, help

[kas1e]

While working on making proper port to some PPC computer, found the last remain issue about big-endian : loading of the .mdl files crashes (in all the other parts Irrlicht seems already support BIG-Endian cpus very well, still )

The file in question to which stack-traces point out are CAnimatedMeshHalfLife.cpp , which didn't have any big-endian fixes as all the other loaders in Irrlicht have. Probabaly that one was added more or less lately, so those who works back in times wasn't worry about adding there ppc fixes.

I start myself to make something, and there is what i got currently (search for "BIG_ENDIAN" ifdefs):

http://kas1e.mikendezign.com/aos4/irrli ... lfLife.cpp

It still crashes, so i probabaly forgot something , or doing something wrong.

I also had to comment out there "const" in few places, so my byteswapping works , but i think it should't cause issues.

Can any expirienced coder have a look at that, maybe found something obviously wrong in my changes , or , maybe something purely forgotten to be byteswapped ?

Also, if anyone willing to help for real with it (even just point me out on what change and what to do remotely, so then i can build it and test on my side), i can pay up to 50$ easy.

Of course those changes can be then merged to trunk once it will works on PPC as well (if CuteAlien doesn't mind of course)

Thanks !

[CuteAlien]

Yeah, different loaders written by different people. And not everyone has a system to test endian-troubles. I think I once had a setup in some VM for basic testing a few years ago ... have to check if I still have that somewhere.

If you want that code in Irrlicht (it should be there) then please don't use SDL_SwapLE32.
Irrlicht seems to have os::Byteswap::byteswap, I haven't worked with that yet myself, but it seems to be used in other loaders.

As for problems - only idea I have is adding printfs for all the numbers and checking if you get same results on both operating systems.