Anaconda fault modes

[Jeremy Katz]

On Sun, 2005-09-18 at 20:08 -0400, G.Wolfe Woodbury wrote:
> I add my voice to Willem's
> 
> Anaconda is a major componenet in getting the system installed, it
> should "Just work"(TM) 

When we get to a release, I agree.  But unfortunately, we need a
development environment too.  And since we depend on a whole system,
that environment is rawhide.  This is part of why, in pre-Fedora days,
rawhide trees weren't even an attempt at being installable.  

> I uderstand that there is major work being done on it now, and I hope
> that the anaconda folks are reading the fedora-test-list where I will be
> posting the results of my continuing attempts to do installs of rawhide.

We're reading, but unfortunately, the mail doesn't do a great deal to
help.  Let me outline the situation of how things generally work:
* rawhide compose starts at 5 am, generally the tree is ready or close
to it by the time we get into the office
* sync up to newer tree, get to work on fixing stuff that's broken and
adding the new stuff.  test as much as is possible with RHupdates
* end of the day rolls around, build a new anaconda package
* go home, rest :)
* read mail/bugzilla about stuff that was fixed earlier in the day :-/
  * ... continued into the next morning.

Not sure how to improve things while still being able to keep moving.
Definitely up for suggestions on ways to help the problem.

Jeremy