On Fri, 13 Feb 2004 23:00:00 +0200, Ville Skyttä wrote:
Further, it would be more "correct" process-wise if the
initial package
submission was the attachment.
Which would also make it easier to find the most recent package in a
ticket. Scrolling up from the bottom to search for URLs in comments, which
announce an updated package, is not ideal.
Then, reviews would be posted by editing
that attachment, setting the status, and binding the reviews clearly to
that particular submission/revision. However, it would result in
reviews being posted as the "edit attachment" comments, ie. again
inlined.
Right now I don't really care for the exact procedure, since working with
an increased number of attachments might turn out to get a bit confusing
at times. Still it would be nice to have a way to locate reviews.
I consider it an unfortunate circumstance, that a few reviewers spend time
on a package which won't be published because it needs another
review. Often, after some time, the packager updates such a package even
after a first publish vote, and the reviewer would need to check that
update again and hope that another person joins to get the second
approval. Fine if fedora.us package queue becomes a source for working
src.rpms, too, but that's not enough.
Some stuff to check:
- Is it possible to create a query that does not match attachment
statuses for obsoleted attachments?
Shouldn't be a problem since attachment statuses can be taken back,
if need be, and queries can be limited on "open" tickets.
- Is the Bugzilla web (and email UI) good enough for implementing
this
or will it cause too much confusion to be worth it?
We probably need to test-drive this a bit. Not as a new policy, but as an
option.
This is already even more complex than the "the review is the
attachment" approach. Dunno...
As a last resort, we might try the REVIEWED keyword, too.
Much more feedback from the target group of these changes is necessary.
--