Testing Fedora - small (?) suggestion.

[Nicolas Mailhot]

Le samedi 11 novembre 2006 à 13:34 -0500, Jesse Keating a écrit :
> On Saturday 11 November 2006 13:26, Nicolas Mailhot wrote:
> > Not if we block only the broken parts. You know, instead of using
> > testers to shame maintainers in fixing broken deps, have the buildsys do
> > it, and only expose sane trees to users (with the "broken" packagesets
> > being held where the buildsys can find them till they're complete)
> 
> Thats quite the complexity to try spinning the tree, find broken deps, drop 
> out packages, try again, rather, rinse repeat.

The yum plugin does it, the auto-dep check does it, I won't say it's
peanuts technically, but it's not rocket science anymore. Even if it
takes an hour of CPU time that's better than making hundreds of testers
waste 15 min.

> Look, rawhide is just a work in progress, a snapshot of what happened the day 
> before.  We can't aways have a completely stable tree. Trying for that is 
> the road to insanity.

No it isn't. A release ago you'd have written forcing every maintainer
to get its buildeps right would never work. The problem is only to get
the right tools in the right place, and try harder.

> This is why we have test releases where we freeze 
> things to get the tree into a sane state.  Could we do more frozen iso 
> releases?  Possibly, depending on the build system we use and how freezes can 
> be handled.

You're not hearing what people are saying. Releasing broken tree and
isos actively discourages testing. When rawhide breaks a lot of testers
will will just pass and wait till anaconda and yum are happy. Others
will try to update nevertheless, spend all their energy manually
workarounding the breakage, and not test anything once the update boots,
because they'll have wasted their free time testing budget. No matter
how you look at it, every period when the tree is broken is a neat loss
on the QA front.

If you want Fedora get the polish it lacks rawhide need to be
installable 90% of the times and FC Test users need to spend their time
on something else than getting the release to boot. Remember "upstream,
upstream upstream". If a problem is not detected early enough to
percolate upstream and get fixed in time for inclusion, we're only doing
testing for other distros. The final release will always reflect the
standard you used in the devel branch, no last-minute test release
effort will change this.

-- 
Nicolas Mailhot