maven-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Benson Margulies <>
Subject Re: The JIRA chocolate box
Date Sun, 19 Jun 2011 11:30:50 GMT
FWIW, I've spent plenty of time on the 'trying to make a test case'
side of the fence and I appreciate how hard it can be.

If someone makes a cogent case that they can't come with a testcase,
then there are possible outcomes:

1: A dev finds the story sufficiently compelling to try to build a
test case or otherwise analyze the problem, soon enough that the
caravan does not move on.
1a: they come up with something and the OP can confirm that it fixes things.
1b: they are stumped or whatever they come with doesn't work.
2: no one in the dev community is willing to tackle.

In short, stating this policy doesn't stop us from trying to be
helpful in the impossible case, I just think that we should be
realistic in declaring failure to avoid recreating the compost heap.

As for github, I don't mind having the email stick to 'attach a zip'.
There are cases where github, or even, 'this svn URL of another Apache
project' allows collaboration with the OP in refining a test case, but
OPs will think of this for themselves.

I don't know that we need a JIRA day; we need everyone willing on dev@
to start digging down the JIRAs pinging.

On Sun, Jun 19, 2011 at 7:02 AM, sebb <> wrote:
> On 19 June 2011 11:52, Dennis Lundberg <> wrote:
>> On 2011-06-19 00:30, Benson Margulies wrote:
>>> I just looked at the 'blocker' issues. We have a variety of very old
>>> JIRAs here. None of the ones I looked at have a self-contained test
>>> case that would can be downloaded, run, and converted to an
>>> integration test, etc.
>> This is one of the major obstacles we have at the Maven project. The
>> sheer amount of old and sometimes really old issues in JIRA. For the
>> last couple of years I've pinged reporters of new issues straight away
>> for test cases, if none was attached to a new issue. But as you say
>> there are lots of old ones that are practically impossible to resolve
>> without some kind of test case.
>> Maybe we should plan a JIRA cleanup day? Let's find a date where some of
>> us on the dev team can work together to clean up old issues. We announce
>> that together with your proposed letter to the users list as well.
>>> What's the policy? My temptation would be to comment on them asking if
>>> the OP is still interested (in some cases, 5 years later), and, if so,
>>> can they come up with a repeatable test case, and if not close as not
>>> a real bug.
>> I've done this kind of thing for plugins in the past. For those I have
>> opted to close the issue as "Incomplete", because it lacks a
>> reproducible test case.
> +1
> This will make it easier to find later if necessary.
> I think it's vital that the resolution distinguish such cases from
> fixed bugs and non-bugs.
>>> I don't mind in some cases doing work to build a test case, but to go
>>> to all this trouble for a bug that was opened about maven 2.0.x, where
>>> it may not be that easy to reconstruct the critical components of the
>>> problem, seems a dubious use of time.
>> Agreed.
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail:
>>> For additional commands, e-mail:
>> --
>> Dennis Lundberg
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail:
>> For additional commands, e-mail:
> ---------------------------------------------------------------------
> To unsubscribe, e-mail:
> For additional commands, e-mail:

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message