On Wed, Oct 14, 2009 at 10:00:07AM -0500, Mike McGrath wrote:
On Wed, 14 Oct 2009, Jesse Keating wrote:
> On Wed, 2009-10-14 at 09:37 -0500, Mike McGrath wrote:
> > On Wed, 14 Oct 2009, Jesse Keating wrote:
> >
> > > On Wed, 2009-10-14 at 09:27 -0500, Mike McGrath wrote:
> > > >
> > > > The problem isn't GLODA and smart folders, it's that we have
no process in
> > > > place to identify and deal with problems like this before it's
too late.
> > >
> > > Aside from updates-testing you mean, where people can test potential
> > > updates and give feedback as to how they work on their systems?
> > >
> >
> > Fat lot of good it's doing.
> >
> > -Mike
> >
>
> And that's a people problem more than a process problem. If nobody
> tests it in updates-testing, then how is the maintainer to know that it
> is problematic? Certainly not solvable with even more repos for testing
> content...
>
You let me know how three people in Fedora can miss a very subtle Firefox
memory leak. How many people would need to use updates testing before the
thunderbird indexing problem is caught? How long would it need to stay
there? In this case updates-testing theory just does not match reality.
The status quo is broken, doing nothing will keep it that way.
I think there are many things we can do within Bodhi & Fedora Community
to better facilitate testing updates. For example, I think we should do
a better job of emphasizing the importance of certain updates in the
queue, especially security updates with little or no karma.
The karma system that I implemented in bodhi hasn't really changed much
over the years, and I think it's probably time to reassess some of the
default options (eg: +3 for marking as stable is probably way too low
for packages like the kernel and thunderbird).
Since there are actually a lot of people who provide feedback in bodhi
(presently 1448 different people have left comments in bodhi since F10),
we obviously are not doing a good enough job of leveraging them.
One way to improve our testing strategy would be to keep adding and
improving the test cases to the wiki:
https://fedoraproject.org/wiki/Category:Test_Cases
We could then point people directly to the appropriate steps for testing
the application, from within bodhi, fedora community, etc...
luke