please stop...

Jeff Spaleta jspaleta at gmail.com
Mon May 17 22:31:46 UTC 2004


On Tue, 18 May 2004 01:10:22 +0300, shmuel siegel
<fedora at shmuelhome.mine.nu> wrote:
> There is nothing wrong with someone expressing saying that it is not a
> good idea to release software that says it will do something but ends up
> trashing an existing setup. In case you didn't notice, Redhat took this
> problem seriously and publicly asked people to describe environments
> that failed. Even our frequent warnings from Jef "the transient middle
> name" Spaleta, never suggested that release software eats babies.

I also never suggested hardware level problems dont make it through
into full releases either.  Sadly, the people reporting the problem
haven't been very specific about mobo and bios specifics. The fact
that one reporter in a comment was able to magically stop seeing the
problem with what sounds like a bios upgrade
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=115980#c11
suggests that this is a very subtle interaction at the hardware/bios
layer and its NOT going to be easily reproducible by developers unless
they can get their hands on systems that are seeing the problem.

What would be really cool...is if someone can try to reproduce the problem
outside of the installer environment, using grub-install in a
controlled manner, both under 2.4 kernel and a 2.6 kernel, using as
much else similar as possible...i am not convinced that this is not
something at the kernel level. I don't claim to fully understand how
grub gets its disk geometry information. But trying to do this in a
controlled way..and figure out if this is limited to the installer
environment...and if this is a 2.6 kernel specific issue...might get
us somewhere.

-jef"test releases eat babies...dual boots eat data... backups have
always been advisable when doing any upgrade or install on a drive
with pre-existing data..and will always be advisable"spaleta





More information about the fedora-test-list mailing list