I've just begun to mess around with clang and llvm, and I'm interested to hear if anyone has tried compiling PCSX2 using it.

C++ support isn't complete, but I think it's worth a try. Not entirely sure how to get it to create a Windows PE but that's a small problem compared to making the code compatible with clang. I can't submit code back to the project (non-compete agreement) but if I get something working it's certainly worth more exploration.

Suggestions/constructive criticisms are appreciated. I know it's likely to incur a severe performance penalty, but I hope with time the advantages of llvm shine through. Or maybe it'll never match the static compilation, I've never heard or read of a bt/recompiler done this way.

Well, LLVM is gcc derivative, so only way to produce a windows binaries is mingw-linux path, which is far from being supported. And henceforth, I don't understand what goal do you assume to reach?

I've found several documents about getting llvm to kick out a Windows executable, but I've never personally tried it. May be impossible to make it then work with the plugins, but as far as I know no one has tried it.

I'm hoping the llvm can (eventually) provide a speed boost to PCSX2. I've tried compiling a very primitive bt/recompiler and saw a nearly 100% increase in speed (to be fair it was a pretty crappy implementation), but I'll be satisfied with coming anywhere close to the current official builds in terms of speed.

I don't want to fork the project or anything, just see if this works. clang/llvm are under constant development, if it provides a benefit it would be a nice alternative in the future. AFAIK llvm can be configured to become a drop-in replacement for GCC, doing that could give a direct comparison.

All of this is useless until I get time to see if PCSX2 is compatible with clang, as C++ support is pretty shaky. This is as much about my education as it is about getting PCSX2 to run, as long as I'm learning about clang/llvm I may as well do useful work