Hey,

Alan Burlison wrote:
> Paul Jakma wrote:
> 
>> Eek, sounds bad alright.
>>
>> What I had in mind was more the ability to refer to external bugs. E.g.:
>>
>> - When creating a bug, to be able to use information from an external
>>     bug as a template for the Solaris bug (e.g. if nothing else, the
>>     subject).
> 
> What happens when the Solaris bug is changed?  How does that propagate
> to the external bug?  What happens when the external bug changes?  How
> does that propagate to the Solaris bug?

Arguably you'd be back at the start with the various restrictions of what can
and can't be propagated according to the changes made.

> This approach won't work, unfortunately.
> 
>> - The ability to link easily to the external bug, ie rather than
>>     having to say "See http://bugzilla.....";, to have a box to enter
>>     the bugid into so that viewers then could be given an easy path to
>>     the external bug (inline viewing of the information or via browser)
> 
> Viewing isn't sufficient, you need to be able to edit both, at the same
> time.  And commit the changes to both, atomically.

I'm not sure you would - I think such a thing could really only work on way eg.
upstream to distribution. Doing it the other way requires some level of care
about exposing sensitive information.

>> - Ability to have certain key fields of the external bug show up in
>>     our bug DB. Bug resolution state particularly.
>>
>>     I.e. If I have a bug in our system which is tracking some external
>>     bug, and the external bug changes to closed/fixed - I'd like to
>>     that to show up.
> 
> And what happens to the external bug if the internal bug is closed?

You could argue 'we don't care'. It should be really up to the distribution to
mark updates in the upstream bug. Yes, it does create a lot of other problems -
lack of bug fixes going back upstream and the possible duplication of effort for
bug fixes between distributions. If you do care and propagate changes upstream
to the external bug tracker, then there's always the possibility that the
reference implementation of whatever code is different to the distribution
specific changes leading to somewhat grumpy developers [1]


Glynn

[1] Been there with GNOME plenty of times before. I'd be interested to see how
    things like Launchpad deal with this
_______________________________________________
tools-discuss mailing list
[email protected]

Reply via email to