Neil said:
> Chris G. Demetriou wrote:
> 
> > Just fine in the sense that it compiles most things, and when you work
> > around a few 'known bugs' (e.g. optimization failures in usr.sbin/sa
> > and usr.sbin/dhcp/server) it'll build the system?
> 
> Yep, just fine in that it _works_ just fine apart from the odd known bug
> (which manifests very rarely). I don't know if it itself builds just
> fine or not.
> 
> Anyway you know the answer to this better than I---why ask me?

A couple of (real) reasons, actually:

(1) you said the 1.3.1 compiler "works just fine."  A legitimate
interpretation (perhaps the only legitimate interpretation 8-) of that
means "with no bugs," and, though I (claim to 8-) have a clue, my
build environment is Really Weird compared to 1.3.1, and so I don't
have absolute trust that what i've seen are real bugs.

(2) the CPU chips in my machines apparently have at least one hardware
glitch, and I don't know what, if any, fraction of problems i've seen
with the compiler to attribute to it.  (Examining binaries, etc.,
makes me think a very small fraction, if any, but if somebody else
sees a different result, that's a very interesting data point!)




cgd