[GRASS5] Dealing with old bug reports: new GRASS 6 bugtracker?

Markus Neteler neteler at itc.it
Fri Jul 22 04:08:26 EDT 2005


On Fri, Jul 22, 2005 at 07:52:13PM +1200, Hamish wrote:
> > just a comment on how to deal with old bug reports in our
> > bugtracker:
> > 
> > In the QGIS project old reports are automatically closed if
> > the reporter doesn't answer a comment in n days:
> 
> The bug doesn't go away even if the the person does.

That's not always true:

- In the posted QGIS case the bug was solved but I forgot to confirm.
  So the auto-closure was correct (otherwise I could reopen it if I
  am not happy yet as user).
- In some/many GRASS cases the developers forget to close in RT or don't
  remember that it was posted there.
- In some cases the problem is solved in later versions, so the bug
  no longer applies if we don't want to backport all features (impossible).

> We can do this for "GRASS doesn't install on my PC, why not?" style
> reports, but I don't think we should do it for segfault-like or wish
> requests.

Agreed for wishes.
Segfaults depend on lots of environmental factors which are often hard/
impossible to reproduce.

 
> > In summary: The 452 bug reports do NOT reflect the current that of
> > GRASS 6. It's simply misleading. 
> > 
> > Other suggestion: We set up a separate bugtracker for GRASS 6.
> 
> I like this idea. Just as the GRASS 5 -> GRASS 6 transistion, only items
> with a sponsor get migrated.
>
> I would not like to have old non-fixed bugs get tagged with the same
> "Resolved" label as fixed bugs.

There is a "stalled" label as well. Maybe newer version of RT provide
even more.

> The bug tracker may contain details
> which might be useful for a user who has found a 5.4 v.digit bug. I know
> the debain bug reports are often reassuring that the error wasn't my
> fault, even if it is old and unfixed.

This requires an RT manager who merges reports etc.

> A lot of the bugs may be 2+ years old, but a lot of the raster modules
> haven't been modified in that long either!

Examples:
- #35 ?
- #241 will never get changed (probably)
- #1035 the same
- #1047 would be nice
- #1063 unlikely
- #1065 unlikely
- #1066 unlikely
- #1232 no idea
 
and so on.


> The bug tracker is not a PR tool.

But it's considered as such a thing by some people...
 
Also: a bug tracker filled with 4xx reports is hard to maintain. At least
personally, I just get frustrated by opening it :-)

> I don't mean to be against the plans, just when the cleanup happens do
> it with care..

Absolutely. That's why I suggested splitting into GRASS 5 RT and GRASS 6 RT.

Markus




More information about the grass-dev mailing list