On Sun, 2007-09-02 at 17:52 +0200, Alberto Passalacqua wrote:
Hello everyone, during a discussion on IRC it emerged that GNOME still has a very high number of blockers and critical bugs. As it can be seen here:
GNOME + Evolution has 26 of them. Plus, from bugzilla it appears that GNOME + GNOME Admin/Platform + Evolution have 201 bugs:
Considering the short time before the final release (~1 month), I think it would be helpful to know the actual situation of the fixes, and if there's enough time to solve all these issues before the release date.
I'll answer the original mail, but I'll collect thoughts related to
other replies here too. For the GNOME bugs in particular at least there
are several issues causing the count to be high and not well maintained:
1) We (as the GNOME maintainers) have not been good at redirecting bugs
upstream. We do have the keyword "should_go_upstream" in use which
indicates what we want to upstream it and resolve it in the opensuse
bugzilla, but we've been slow to do this though and its something we've
not publicized and that was probably wrong.
2) We have not been good at grading bug severity, generally severity has
been left at whatever the filer put it, although Blocker/Critical/Major
have undergone more scrutiny in 10.3 at least.
Further complicating this is that we (as a project) do not define
severity well for instance from what I could find:
http://en.opensuse.org/Bug_Reporting_FAQ
http://en.opensuse.org/Submitting_Bug_Reports
Only one mentions severity but its a brief reference to blocker. The
ones on the Novell bugzilla are also weak:
https://bugzilla.novell.com/page.cgi?id=fields.html#bug_severity
3) We (as the GNOME maintainers who work for Novell) have traditionally
not focused on openSUSE much, rather spending most of our time on SLED.
This improved for 10.3 and well continue to improve, but we lagged a
little more than we thought due to SP1 of SLED 10.
4) As some one else mentioned, you almost never close out all the bugs
for release
Now, not one to simply list problems with out a solution, here on some
positives:
1) I've been monitoring the count, its been trending down over the last
three weeks (including major and above)
2) There are a number of easy packaging related things
(bnc-team-gnome-build) we can pick off before 10.3
3) Mark Gordon a Novell QA person has been able to spend a lot more time
triaging and moving bugs upstream over the last several weeks
4) Of the criticals, only 3 have not been active (generally activity
means they are getting attention on the way to a fix) in the last week
and they are:
https://bugzilla.novell.com/show_bug.cgi?id=299331
https://bugzilla.novell.com/show_bug.cgi?id=299005
Both being worked on by hpj as part of a larger set of help fixes
(rarian integration), its just a big mess atm because we have to tie
this help system into the suse help system which is based on how kde
discovers help files and our older patches broke big time. We should
look at the for 10.4
Thats good and all , but it doesn't go all the way of course -
Gary Ekker has been getting some better bug criteria created for
understanding when we should upstream, wontfix, invalid, etc. Its for
both SLED and openSUSE, we did a test run last week and it worked
reasonably well. I think Gary would like to publish these soon for
GNOME at least, just a bit of hesitation to push them out at this point
in the release cycle. Maybe AJ/Coolo/Christoph could take a quick look
first and they can form a basis for discussion across the project
(although maybe upstreaming differs between kernel and GNOME).
Finally, there has been a move to standardize severity definitions
internally at Novell, we should probably see about publicizing those or
consciously have different (more standard open source definitions) for
those. I have some definition thats Gary, Thorsten K, Mihnea and I did
a few months a go that could be a good start once they are un-SLEDified
if anyone is interested.
(I suppose these last two things should migrate to an opensuse-project
thread.)
-JP
--
JP Rosevear