Mailinglist Archive: opensuse-project (146 mails)

[Ruediger Oertel <ro@xxxxxxx>]

On Wednesday 25 February 2009 22:22:18 jdd wrote:
> Gerald Pfeifer a écrit :
> > On Wed, 25 Feb 2009, Eric Springer wrote:
> > > In short, try run factory as your main distro for a month -- and the 
> > > problems testers have should be pretty obvious =)
> >
> > I've been doing this for half of last year :-), including full 
> > distribution upgrades in hotel rooms overnight, and in general I
> > found this surprisingly stable.
>
> is that possible right now? I failed to install factory from network
> with the factory mini cd right now (on virtualbox)

I think this is exactly the difference between the two attempts of installing.
As far as I understand Gerald (and this is exactly what I'm doing), we are
running factory all the time. Which means in detail: we installed at some
point in time (for example the machine I'm writing this on saw the last install
somewhen around the 11.0 release) and since then I usually run either "zypper 
up"
or "zypper dup" roughly twice a week to keep up. If this fails, or the result 
sounds
too scary (like zypper dup trying to change the architecture of 100+ packages),
I call up "yast2 sw_single" select the repositories, a package, context menu
"all in this list" and then "update if newer version available" and look at the 
problems.

This is a completely different issue than trying to install a system or image 
from scratch.
Any severe bug in either the instsys or any critical yast2 configuration module 
can block
the latter case, while you would not even see it in the former case.

One is testing the system as such (my most frequented apps still run, my 
desktop still
starts up during login, and so on) versus tesing the installation workflow and 
getting a running
system. Both methods are very valuable, but substantially different and I can 
easily see
the "install from scratch" method being broken much more frequently.

For example if a kernel is updated and some out-of-tree module used on my 
machine does
not build any longer, then the "update-the-system" variant will most likely end 
up not updating
the kernel at all and keep that for the next time when that package (the 
foo-kmp) has been fixed.

The "install-from-scratch" variant would end up with some hardware not working 
at all.
The same is probably true for other subtrees of packages where one library was 
updated
with a set of packages failing to build because of that and still requiring the 
old library name.

(not trying to excuse any failure, just trying to explain here ...)
-- 
with kind regards (mit freundlichem Grinsen),
   Ruediger Oertel (ro@xxxxxxxxxx,ro@xxxxxxx,bugfinder@xxxxxxxxxxx)
----------------------------------------------------------------------
Linux MacBookRudi 2.6.29-rc5-1-default #1 SMP 2009-02-16 17:18:44 +0100 x86_64
Key fingerprint = 17DC 6553 86A7 384B 53C5  CA5C 3CE4 F2E7 23F2 B417
SUSE LINUX Products GmbH,  GF: Markus Rex,   HRB 16746 (AG Nürnberg)

--
To unsubscribe, e-mail: opensuse-project+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: opensuse-project+help@xxxxxxxxxxxx