Christoffer Lerno
Christoffer Lerno
I've made a commit #571 which addresses the int issues. It's still a bit unclear to me how I want to support float16, float128 and int128 on platforms that don't...
Those limits are in math.c3 already.
float.min now returns -float.max
Yes, no support for float128 yet.
This should now work. Can you test?
It resolves that particular error, but in the linking step it now ends up with: ``` /usr/lib/x86_64-linux-gnu/libtinfo.so /usr/lib/llvm-15/lib/libLLVMDemangle.a && : /usr/bin/ld: /usr/lib/llvm-15/lib/libLLVMAArch64AsmParser.a: error adding symbols: archive has no index; run...
@sylvestre Regarding LTO, I previously had problems with homebrew enabling LTO in their llvm15 package, as this prevented linking with anything except the exact same Clang version they used. So...
They then switched to thinLTO which also didn't work :(
Yes, the brew issue was resolved.
No, while the workaround with bolt-15 now seems to be ok, I don't think CMake for LLVM targets should break if bolt-15 isn't imported.