reporting bugs upstream : nothing on the wiki?
sanjay.ankur at gmail.com
Tue Jul 13 08:36:56 UTC 2010
On Mon, 2010-07-12 at 19:44 +0000, "Jóhann B. Guðmundsson" wrote:
> On 07/12/2010 07:24 PM, Adam Williamson wrote:
> > It's a topic that's currently under discussion on -devel. In practice
> > I'd say it varies. Some reporters certainly are willing to make good
> > upstream reports. Some aren't, indeed, but I think Ankur's probably
> > right that we can add value to the ecosystem (urgh, excuse me while I
> > shoot myself for that one) by providing good instructions for those who
> > _are_ willing to go the extra mile.
> Those instruction for those components should then be written and
> maintained by the maintainer(s) of that component.
This is an interesting idea. I don't think the maintainers themselves
would have the time to do this. Can this be passed on to the mentors, to
pass on as an activity for new Ambassadors ;) ? They'll learn too in the
process, and might end up co-maintaining some packages (which is a good
thing!). It would be a good idea to have as many people as possible
exposed to the bug reporting process. Starting with new Ambassadors is
some what going down to the roots. Comments?
> That means he/they will need to create step by step a.k.a spoon feeding
> instruction for. . .
> ( keep in mind here the this may be an individual that has no technical
> knowledge what so ever and took the time to file his/her first report.
> The response to that first report will be the deciding factor if that
> individual continues to file reports in the future )
> How to create an account in upstream bug tracker ( not all upstream bug
> trackers are mozilla bugzilla ).
Generic instructions for this are available everywhere. This is simple
and hardly takes a minute. (I don't know how many bugzillas I've joined
> How to open up a terminal and compile the component from scratch from
> upstream repository encase upstream rejects the report based on what's
> package and shipped in Fedora or requires the reporter to recreate the
> failure using the upstream bits.
Don't we have a policy of staying close to upstream? Are there a lot of
differences in the upstream package and what fedora ships? I maintain a
few packages, and they're all same as the upstream source. (just
This depends on the bug and cannot be generalized. Bugs that are
reproducible on all machines will not require this. Upstream will handle
those (I'm sure they'll be happy to do it themselves than wait for the
reporter to get back to them , if ever). Machine specific bugs have no
other solution. If upstream can't reproduce it, it has to rely on the
reporter for feedback. The maintainer can't do much too here.
Upstreams interested in getting good bug reports do keep a good
documentation of how to go about it. If they don't, we can't do much,
it's really their loss. What I'm saying is that we encourage users to go
all the way to upstream and report bugs rather than stopping at our
bugzilla. As Jóhann already mentioned, rather forcefully, this is a
difficult to achieve goal. Even partially achieving it would be an
More information about the test