On Wed, Oct 11, 2000 at 09:17:39AM -0400, Robin S. Socha wrote:
> * Peter Green <[EMAIL PROTECTED]> [001011 08:52]:
> > Okay, aside from the less helpful remarks you've received, I wouldn't
> > suggest running anything after RedHat 6.1. I've encountered some weird
> > problems with 6.2,
> As per usual, you had to install the errata. Wasn't as bad as 6.1,
> though.
I'd have to agree. I have about 15 RH boxes, with 5-6 running 6.2,
including 2 of them running qmail with no problems so far.
> > and IIRC 7.0 shipped with a highly experimental compiler and glibc.
> > Blech. (I just saw on another mailing list where something broke
> > because of it.)
I have yet to see(or hear) about anything that failed to compile because
of the compiler. The biggest reason stuff fails to compile is failure to
follow ISO C(and C++). If consider only C, it doesn't really matter which
compiler you ship. If you consider C++, it depends. "2.96" has some new
features that RH wanted to take advantage of, and the different versions
aren't binary compatible anyway, so why not use that one that gives you
the features you want?
Both compiler and glibc works fine. Look at it this
way. Everything(except kernel 2.2.17 and kde1) in 7.0 was compiled with
the experimental compiler, and everything seem to work just fine for
me. I even compiled 2.4.0-test9(which I have running on my desktop
box) with the new compiler and it works fine too.
> Probably the lusers pathetic attempt at emulating intelligence. kgcc is
> certainly one of RH's weirder ideas, but it's fully documented. Don't
kgcc is actually not RH's idea, it's Conectiva Linux.
> > We have about 20 RH servers here, run qmail on RH 6.1 w/o a
> > problem...but they are making some stupid decisions, IMO.
I haven't really seen anything in RH 7.0 that I find stupid yet. Perhaps
I would do certain things different, but that doesn't make theirs stupid.
Obqmail: I would probably run qmail on RH 6.2 for now, and put 7.0 under
testing when you get time.
Ole Gjerde
