[Tracker-discuss] type field, specifically 'crash', and proper setting of 'versions'

[Brett Cannon]

On Tue, Mar 24, 2009 at 18:25, R. David Murray <rdmurray at bitdance.com>wrote:

> On Tue, 24 Mar 2009 at 21:17, Daniel (ajax) Diniz wrote:
>
>> Brett Cannon wrote:
>>
>>> Also, what "should" be selected in the versions field?  All versions in
>>>> which the bug appears?  Only versions in which the bug might get
>>>> fixed?  The version in which it was reported plus any later versions
>>>> in which it is still broken?
>>>>
>>>
>>> Any version it is known to be broken in. So if it was reported in 2.5 and
>>> it
>>> still broken in trunk then 2.5, 2.6, and 2.7 should all be flagged. It
>>> lets
>>> people who are using old versions find out what they might have to watch
>>> out
>>> for since we are not about to fix anything for 2.5 that is not
>>> security-related.
>>>
>>
>> Oopsie, I've applied a different rule to a couple of hundred of issues :)
>>
>> I'll use this rule from now on and eventually fix the wrong changes.
>> IIRC, my wrong interpretation is somewhat widespread, so it might be
>> interesting to add help/descriptions to the versions input.
>>
>> Some doubts:
>>  For a bug filled against e.g. 2.4 that was confirmed then but not
>> yet in any new versions, should we add current versions to denote they
>> need checking?
>>  For bugs filled against e.g. 2.5 that were never confirmed, should
>> we add the current version? And should we try to verify it for the old
>> one?
>>  For bugs that were confirmed for e.g. 2.3 but don't have that
>> version set, should we be able to add it?
>>
>
> IMO, given what Brett said, we should only add ones for which we can
> confirm the bug exists, so if there's no test then it would only be
> the one the submitter reported, with the stage set to test required.
> If there's a test, then run it.  I would imagine we could assume that
> if the bug exists in trunk it exists in all versions in between, so we'd
> only need to run two tests in most cases (trunk and py3k).
>

Exactly.

-Brett
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/tracker-discuss/attachments/20090324/69254c99/attachment-0001.htm>