Just a bit of perspective to add to the discussion with respect to
major changes in a distribution.
I started at Red Hat when 2.1 hit the streets. 2.0 was still going like
mad (the first ELF release of Red Hat) and there were complaints about
getting a.out binaries to work properly.
2.1 went better, and was out there for a while when it was decided
(pretty much by accident) that it was time for a new release, and
3.0.3 was born. (Keep in mind that at the time the RH development
office consisted of 4 people with limited resources, we tested as
best we could but there wasn't a formalized procedure.) With the
heavy networking that we used internally there were some glitches that
folks reported that we couldn't replicate. And we were all running
3.0.3 systems (with some modified 2.1 systems tossed in for variety :-)
If I remember correctly the upgrade from libc 5.0.9 to 5.2 wreaked a bit
of havoc similar to the move from libc5 to glibc (at the time we were
already discussing the glibc project and if/when it might be ready to
include in the distribution) Other than that it was a smooth release
and went well overall. (even if the box looked like a color chart
from Duron :-)
Then 4.0 was released and the move to the 2.0 kernel posed a challenge.
More people were irritated that they had to *reboot* their machines to
upgrade than anything else. (None of them could come up with a better
solution for upgrading the running kernel. :-) The testing for 4.0
was more formalized than ever before, but things *still* got overlooked.
4.1 was better, 4.2 was/is an excellent iteration (I put it and 3.0.3
in a class together)
Now, we haven't seen any real problems with 5.0 (I upgraded my
workstation the day I got the CD) other than those that have been
addressed. Does that mean they don't exist? No. Does
that mean we won't find something broken? No. Does the fact that
problems do exist mean Wangsmo & Co. are spending their time playing
Quake when they ought to be testing? Not likely.
(Maybe just a little ? :-) If you take a step back and look at the
complexity of the O/S it soon becomes apparent that for a few people
(or even a few thousand people) to test everything is impossible. RH 5.0
went through several beta releases (3-4?) and lots of problems were
resolved.
Am I biased? Admittedly so. Will Red Hat ever put out a distro that
is bug free? I'd wager my yearly salary that no one could keep a viable
release schedule for an equivalent product and not have similar results.
This message is considerably longer than I intended, so please accept
my aplogies. My intent is to offer a unique perspective on this topic
so perhaps others could better evaluate the current state of affairs.
Regards,
--Kit
--
--
Kit Cosper Linux Hardware Solutions, Inc. [EMAIL PROTECTED]
System solutions you can count on, system solutions that work.
1-888-LINUX-HW
Now offering Digital Alpha systems with up to 4 MB SRAM cache.
--
PLEASE read the Red Hat FAQ, Tips, Errata and the MAILING LIST ARCHIVES!
http://www.redhat.com/RedHat-FAQ /RedHat-Errata /RedHat-Tips /mailing-lists
To unsubscribe: mail [EMAIL PROTECTED] with
"unsubscribe" as the Subject.
