A word on Trac ticket submissions...

[Scott Webster]

On Tue, Jun 22, 2010 at 12:56 PM, Ryan Schmidt <ryandesign at macports.org> wrote:
>
> On Jun 22, 2010, at 08:08, Jeff Singleton wrote:
>
>> How about configuring Trac with Drop Down selections for maintainers,
>
> We had a drop-down for maintainers. It slowed Trac to a crawl. We removed the drop-down. Most people seem to manage just fine with the text field.

I think this is likely the most common "mistake" people make when
submitting tickets (failing to cc the maintainer).  Perhaps we can add
a little comment next to the cc box that reminds people to do so and
tells them how to find the maintainer address.  I'll try to take a
look at the form and see if I can figure out if this or any other
trivial changes would work.

>> And when the port is broken due to upstream issues, then it should be marked broken
>
> In what way do you suggest it should be marked? We don't currently have an established marking mechanism for that.
>

Possibly we could simply adopt the practice of adding an "upstream"
tag to tickets that involve upstream issues.  One thing I'm not
entirely clear on is what to do when there is an upstream problem that
we resolve with a non-ideal workaround (such as disabling MMX on
#20906).  Do we close the ticket because the problem has gone away, or
leave it open because we should get rid of the work around when the
upstream issue is solved.  I think probably leaving it open is wise.

>> and force a working version to be used instead or a note stating what the issue is.
>
> IMHO the notes stating what the issues are are called Trac tickets.
>

Possibly we could improve the options for people who want to look up
problems with a specific port.  Some ideas are: add a note when the
build fails suggesting looking up tickets, possibly even with a URL
which knows the correct port to search for, eg.
http://trac.macports.org/report/16?PORT=babl  ... we could also likely
improve the link on the trac main page to "Tickets filed against a
specific port" which currently results in something that almost
appears to be an error message.

I might look into some of these things if I get around to it...

Scott