X-pop3-spooler: POP3MAIL 2.1.0 b 4 980420 -bs-
Message-ID: <19980915172355.36334@cerebro.laendle>
Date: Tue, 15 Sep 1998 17:23:55 +0200
From: Marc Lehmann <pcg AT goof DOT com>
Cc: beastium <beastium-list AT Desk DOT nl>
Subject: Re: libc-5.4.22?!?
References: <19980914231210 DOT 00535 AT cerebro DOT laendle> <199809150935 DOT CAA03343 AT Midnight DOT Hacking DOT in DOT the DOT land DOT of DOT Kalifornia DOT com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
In-Reply-To: <199809150935.CAA03343@Midnight.Hacking.in.the.land.of.Kalifornia.com>; from david on Tue, Sep 15, 1998 at 02:35:25AM -0700
X-Operating-System: Linux version 2.1.120 (root AT cerebro) (gcc version pgcc-2.91.57 19980901 (egcs-1.1 release)) 
Status: RO
Content-Length: 2455
Lines: 51

On Tue, Sep 15, 1998 at 02:35:25AM -0700, david wrote:
> > don't want to take their customers to the "experimental" 5.4 releases.
> 
> I would hardly consider 5.4.46 as 'experimental' :)

Officially, it is. For example, linux 2.0 does run much less reliably on my
machine than, say, 2.1.120. Still 2.1 is experimental and 2.0 is not.

Redhat and other distributors have that (understandable) attitude of
having the programs work that have been thouroughly tested, rather than
new, unproven ones.

> Redhat is thereby making things inherently incompatible.  Developers
> increase version numbers on their [libc5] program for a reason.  Making

Developers can increase the version number of their programs at will. There
are (to my knowledge) no user-visible changes in 5.4 as compared to 5.3,
just a "better" internal structure and working NIS and better NLS. The
reason the pgcc binary requires it is simply because it was compiled using
it. Nothing in pgcc depends on 5.4, just the binary does. Same goes with
almost all programs around (except maybe yputils ;)

> several patches and not increasing the version number leads to
> confusion...is it or isn't it fixed for problem xyz?

I object to redhat's practise as well. But apperently they receive the bug
reports, and decided 5.3 is better for them than 5.4.

> These people are sometimes hard to convince to upgrade.  Same goes for

"never touch a running system" ;)

> people with libc5.  If it's working perfectly for them, there is no reason
> why they should upgrade and they'll get a little hot around the collar if
> you drop support for their system.
> 
> But we're getting a bit off topic here.  I wish Redhat would supply
> current versions of software.  They were quite willing to release a
> distribution with a pre kernel on it, I'm sure a library that has been in
> use for many many months with no complaints is certainly valid.

How do you think there are no complaints? There are quite a number of (new)
bugs in 5.4, it simply isn't supported anymore.

      -----==-                                              |
      ----==-- _                                            |
      ---==---(_)__  __ ____  __       Marc Lehmann       +--
      --==---/ / _ \/ // /\ \/ /       pcg AT goof DOT com       |e|
      -=====/_/_//_/\_,_/ /_/\_\                          --+
    The choice of a GNU generation                        |
                                                          |