changes are needed, we need keep moving

Eric Rostetter rostetter at mail.utexas.edu
Fri Jun 3 02:36:35 UTC 2005


I've googled on FL timeouts, and it looks like at one time we said:

 If a similar update is available for multiple similar OS versions (e.g. for
 both RHL 7.2 and RHL 7.3), and has passed the publish criteria for one OS
 version but not for the other, the second OS version may be released after a
 timeout period if no one has tested it. This prevents updates from being
 published for one OS version due to lack of testers when it has passed on other
 similar OS versions and is believed to therefore be safe. Such releases are at
 the discretion of the Fedora Legacy package publishers.

Now that surely applied to "PUBLISH" and I think it was meant to apply
to "VERIFY" also.  But it lacks details (what time period for the timeout,
etc).

>From Jesse Keating on 8-Feb-2004:

 Just needs some more people to look at it. Seems a few packages need either
 7.2 or 8.0 or both QA before we can do anything with them. I'm going to
 continue with my policy of timeout on testing, and just push them if they've
 gotten good 7.3 feedback. If something breaks, thats really sucky, but perhaps
 it'll motivate the community to kick in some 7.2/8.0 love. It was the
 community that persuaded me to include 7.2/8.0 in the first place.

Also got several bugzilla hits which say:

  Pushed to updates-testing due to QA timeout.

So, in summary, I'd surely support something that said "if we've met the
publish/verify status for one or more OS releases of a package, but not
for other similar OS releases of a package, they should all be published
after a timeout period of N days." For this we need to define two things
yet: how many days (N), and what OS versions are similar (and this could
vary by package; sometimes the package may be similar between OS versions
even if the OS versions are not so close).

Comments?

-- 
Eric Rostetter




More information about the fedora-legacy-list mailing list