buildworld failure

[Doug Barton]

On Fri, 29 Aug 2003, Sheldon Hearn wrote:

> Okay, you philosophize while the rest of us follow the advice of the
> folks who have a good understanding of gcc's optimizer. :-)

Not to be disagreeable, but the gcc developers seem to think that -O2
should "always" produce better code, and "never" produce errors. This is
straight from the mouth of the husband of one of my co-workers, who
works for redhat hacking gcc.

If you have a reproducable case where correct C code produces bad
objects, or fails to compile using -O2, the gcc folks want to hear about
it.

Doug

-- 

    This .signature sanitized for your protection