On Tue, Apr 13, 2004 at 09:00:01AM +0200, Adrian 'Dagurashibanipal' von Bidder wrote: > On Friday 09 April 2004 22.24, William Ballard wrote: > > On Fri, Apr 09, 2004 at 03:15:20PM -0500, elijah wright wrote: > > > in my experience compiling stuff with -O3 just means that people on > > > other architectures (where GCC may do odd things) will eventually > > > probably file bugs on your package that can be fixed by moving back > > > to -O2. > > I know how tricky these issues are, but it sounds like the bugs > > should more appropriately be filed against gcc on archs. where -O3 > > breaks. Unless it's squirrelly and it's really not broken on those > > architectures, but the *app* is somehow doing something wrong. > Of course they are gcc bugs and should be duly reported. The question > is: should package x (working with -O2, not working with -O3 > - -fspecially-optimised) be buggy (often FTBFS or RC) because of a gcc > problem when a really trivial workaround is available? I think not, > especially in the usual case where the code is not performance critical > at all. It would still be a bug to turn on -O3 optimization when you don't know what you're doing (which is to say, in the vast majority of cases), whether or not it actually trips a compiler bug in the process. -- Steve Langasek postmodern programmer
Attachment:
signature.asc
Description: Digital signature