Chris, > 2. The debugging tests don't work on cygwin, because apparently debug > info doesn't work there yet. I won't care much about debug information for mingw32/cygwin for now. The reason is simple: tools uses stabs there and, even worse, all runtime libraries are compiled with stabs. Mixing stabs & dwarf makes tools crazy, so I'd prefer to wait, until Cygwin/Mingw32 will turn to DWARF. AFAIK, for mingw32 it's planned for gcc 4.3. So, things "work in general", but not too good due to problems independent from LLVM :)
> 3. There are some portability issues with alloca, as you noticed. This needs to be investigated. "_alloca" is in libgcc and, in fact is compiled from the same source for both cygwin & mingw. It should be also exported from cygwin1.dll, but I don't know, what will be the name in this case (Aaron, could you please investigate?) > 4. There is a small smattering of real bugs, such as llc targeting > cygwin/x86-64. We don't support this target yet, but should fix the > bugs as we find them. I think we should add suitable target triple (not just mark march=x86-64) to that test, so, it won't fail. -- With best regards, Anton Korobeynikov. Faculty of Mathematics & Mechanics, Saint Petersburg State University. _______________________________________________ llvm-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
