Bug #372011 (or: how we could help with anaconda beta tests)

[Andre Costa]

Hi folks,

bug #372011 [https://bugzilla.redhat.com/show_bug.cgi?id=372011] is
hosting a heated debate about anaconda & F8, how this dreadful bug has
been hurting F8 reputation and so on (I'm one of the poor souls that
have been affected by it -- and have been saved by patched .img
provided by Jeremy Katz
[http://katzj.fedorapeople.org/updates-f8-yumloop.img]).

One thing that's clear is that anaconda QA missed some key spots, and
also that we (users) didn't help much on the process, allowing the bugs
to remain hidden until the version was officially released, which led to
a lot of stress among users and developers.

I would really like to participate more during the beta stage of new
Fedora versions. However, as I stated on Comment #97
[https://bugzilla.redhat.com/show_bug.cgi?id=372011#c97] I can't really
install beta versions on my system at home, I use it daily, including
for work.

But, AFAICS the other aspects of the installation process are pretty
much stable and independent from the current installation (language
selection, keyboard selection etc.), and the critical step for an
upgrade is dependency solving / package selection.

So, what if the developers provided early access to this particular
part of anaconda only? I mean, in read-only mode, it would just gather
information about the packages currently installed and confirm if it
would be able to handle an upgrade on a "real" installation scenario?
It could for instance stop right after depsolve and show some
statistics.

Believe me, if I was sure that I could test anaconda in read-only mode I
would gladly do it, at any step before the official release. Chances
are that test coverage would improve considerably, and no installation
would be touched during this process.

Does this make any sense to anyone? Would this help? Is it already
possible somehow?

Regards,

Andre