Neil Conway wrote:
> On Wed, 2004-10-20 at 02:07, Tom Lane wrote:
> > Looks reasonable to me. Just one comment: should the
> > -fno-strict-aliasing probe be inside the "if test "$GCC" = yes" part?
> > It effectively was in the original.
>
> Yeah, makes sense. Patch applied with this fix and Peter's suggested
> improvement to a comment.
>
> > > BTW, since we're on the topic of compiler options, is there a reason we
> > > don't use -g3 with GCC when --enable-debug is specified? It seems worth
> > > using to me.
> >
> > How much does it bloat the executable?
>
> Quite a bit, as it turns out: with CFLAGS="-O2 -g, "postgres" is 7077111
> bytes. With CFLAGS="-O2 -g3", "postgres" is 51227279 bytes. So there is
> quite an increase -- I guess we had better stick with -g then.
Yes, stay with -g. As I remember the more verbose -g information isn't
usable by most debug tools anyway.
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073
