I just found another difference with attachments:
Whereas a comment is emailed to the other people on the CC
list/reporter/etc, the attachment is not. So the email that is sent has
very little value.
I probably will miss the details most when an updated package is
released with the URL listed in an attachment as I'll have to hit
bugzilla to find the URL instead of copying from the email and using
wget to retrieve it.
I also find it can be useful to get emails showing me what things other
people look for in QA (non-showstoppers, what information is relevant to
a QA report, etc.)
On Mon, 2004-02-16 at 13:37, Michael Schwendt wrote:
On Mon, 16 Feb 2004 12:31:30 -0500, Toshio wrote:
> On Sat, 2004-02-14 at 08:06, Ville Skyttä wrote:
> > > Much more feedback from the target group of these changes is necessary.
> > Yep. But let's start experimenting with this stuff to get some hands-on
> > experience how it feels.
> Am I the target group? :-)
> I just QA'd a package and tried out giving
> it a first-review attachment. I think it's significantly different than
> the current approach and marginally more work.
Thanks for trying! I've noticed, too, that creating an attachment and
changing the bugzilla keywords line are separate steps. And someone who
submitted the second-review would need three steps (attach, change status,
change keywords). Personally, I wouldn't mind the extra step for setting
the attachment status. But that's why feedback is needed.
E.g. I like how attachments solve the problem of invalidated GPG
signatures (I haven't had any of those myself, though).
> I think a new keyword is a simpler solution that gets us the bare
> essentials of what we want.
Votes on what keyword to try are still taken. REVIEWED has been
preferred, it seems.
I've thought about another keyword like ON_QA as a silent indicator from
someone adding himself to the "Cc" field that he's planned to review the
package. Of course, a short comment would be sufficient, albeit not
Btw, I think there ought to be more communication between packagers and
reviewers anyway, so a packager doesn't apply severe modifications while a
reviewer is checking the current release.