On Thu, 2002-05-23 at 19:06, Glenn McGrath wrote: > I dont think -march=<cpu> should ever produce a bad binary, -march=athlon and -march=pentiumpro have made a binary of mine behave differently. It was my fault; I was (accidentally) depending on undefined behavior. Interestingly, it worked on FreeBSD, PowerPC, Alpha, and i386 --- as long as I didn't use the wrong -march... Didn't work on Darwin, though. Everything's fine now that I fixed my undefined behavior, but it _could_ break things. Probably isn't to often an occurance, and would be picked up more by people using different compiler versions. Or the upteen different architectures.
Attachment:
signature.asc
Description: This is a digitally signed message part