On Thu, 8 Jul 2010 19:47:41 -0400
Jeff Garzik <jgarzik(a)pobox.com> wrote:
Cute re-ordering of events, there. No, after repeated experiences
with seeking reviews, including this most recent one mentioned
elsewhere on this list, and seeing others on this list repeating
review requests, I was inspired to poke around to see why responses
were so uneven.
IMHO, lack of reviewer manpower and some people dumping large numbers
of reviews into the queue without contributing reviews back.
Looking at the process with fresh eyes, starting from
https://fedoraproject.org/wiki/PackageReviewProcess and moving to
https://fedoraproject.org/wiki/PackageMaintainers/ReviewRequests one
sees a chaotic mess of package reviews, both assigned and unassigned,
not really moving forward at all. Looking closely, you see a lot of
packages that seem of worth, but that set is crowded by review
requests for ancient packages like redhat-menus or kernel.
Yeah, as mentioned, try:
http://fedoraproject.org/PackageReviewStatus/
instead. I think I have changed the wiki to point to this instead, but
if you spot it still listed, please change it.
In an ideal world, every package in fedora/devel would get a full
package re-review prior to each release. But with finite resources
limiting that, it seemed to me that triaging long-dead bugs for
long-merged packages was a reasonable and helpful thing for the Fedora
project. By all appearances, nobody else was bothering with these
things after several years went by.
Yeah, I agree that ideally we would be able to re-review existing
packages, but sadly, the manpower is just not there currently.
I don't think triaging merge reviews is a good idea though. They aren't
harming anyone, and slowly people are doing them, so why not let them
stay?
If people want these obviously unloved, ignored review requests --
not
even an rpmlint or ping in many cases -- to stick around, that's fine
with me. I thought I was being helpful, but easy enough to leave
things alone as well.
Thanks. I would like for us to come up with some kind of plan for
these, but nothing is decided yet. It's best IMHO to leave them open
for now.
My hail review was proceeding, and I wanted to make the process a
bit
easier for the -next- person wanting a review. Apologies for the
ruffled feathers.
No worries.
The process itself is intimidating, because the wikis demand that a
prospective reviewer wade into a completely unorganized swamp (BZ URLs
linked-to from above URLs), hundreds of review requests, with next to
zero information about where one's review would be most helpful. To
an outsider, it must seem like quite a mess, with completely unknown
chances for success/failure.
Yeah, we could look at organizing some more for sure. There was a
"Package review sig" that tried to start up at one point, but as far as
I know it never got anywhere.
Perhaps things like using priority/severity would be nice... you could
then mark reviews that are needed for new features as high, or things
that are leaf nodes as low, etc.
Anyhow, more things to try and work on. ;)
kevin