@FioraAeterna care to chime in here? http://reviews.llvm.org/D18643
@stephentyrone @whitequark @FioraAeterna From a standpoint of conforming to current standards I think it's the only one.
-
-
@RichFelker@whitequark@FioraAeterna From that standpoint we need to continue supporting math-errno. I am uninterested in that world. -
@stephentyrone@whitequark@FioraAeterna No, standards only require either fenv exception flags or errno, not both. errno can be dropped. -
@RichFelker@whitequark@FioraAeterna For stdlib implementors. For users and compilers (targeting arbitrary stdlibs), no. -
@RichFelker@whitequark@FioraAeterna while I have a personal interest in life being easy for stdlibs, that's the least value of the three. -
@stephentyrone@whitequark@FioraAeterna For stdlib implementors, supporting softfloat fenv is more work than errno. -
@RichFelker@whitequark@FioraAeterna I agree, and both are abominations, which is why I want to tear the whole thing down. -
@stephentyrone@whitequark@FioraAeterna For fenv there's a standards-conforming way to say "don't need it":#pragma STDC FENV_ACCESS OFF -
@stephentyrone@whitequark@FioraAeterna Then, if the compiler can assume errno is not used, it can hoist ALL math functions at will.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

