On 23 November 2014 at 19:48, Kristian Rosenvold < kristian.rosenv...@gmail.com> wrote:
> I think this is a very good idea. But I have seen this mis-used a few > times in other projects, and I think we want to avoid this scenario: > > There are some bugs that have very well written bug reports with > detailed descriptions on reproduction and/or quite a few watchers too. > I've seen this "rule" misused; some committer comes along and requests > some kind of additional information - sometimes the request can border > on the silly and maybe the committer does not really even understand > the problem (half of us really don't half the time - that's just the > way triaging works sometimes). So the issue gets closed because the > people who are familiar with the bug think it's a silly request to > make (and they might not even know its a committer making it - we have > no special distinctions in jira) and we basically piss off users. > > So I think it's a good strategy for half-baked, potentially old and > half-clear issues (which there are a LOT of!). A hundred years ago > someone taught me that if one person reports something it can be > ignored, but when the second person reports it it's most likely a bug. > So any issue with just one "user" watcher should be pretty harmless ( > I may sometimes watch issues but I really don't want to be counted as > a watcher...) > > I just fixed/triaged well over a hundred bugs in maven-assembly-plugin > and out of the 56 remaining bugs there's probably 20-30 I'd want to > close this way. > > I was about to create some heuristic about when to be careful with > such a rule, but I basically changed my mind :) > > Actually maybe we should just say that after >30 days, we add an > additional message > > "this issue will be closed as incomplete in 10 days unless the > requested information is supplied" or a response seeking clarification on the exact information required, etc > ; just to give a clear indication > that we mean it. A well defined process is much more important than > "hasty" closing. And if a different committer thinks the issue should > be kept open, it's all ok to say so in the issue. > > I am also slightly sceptical of carpet-bombing jira with this stuff; > once we request more test data we're also giving the expectation that > someone /will/ be looking at the additional data that the user has > supplied. So I would be expecting whoever triages with this method to > also be willing to do at least some followup...? > Yep, I remember doing something like this for core or surefire a while ago... Certainly I think in the drive for 3.2.1 I put a fair bit of effort trying to triage the bug list > > But this is overall just details on how to make this good; I'm +1 on > the proposal. > > Kristian > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org > For additional commands, e-mail: dev-h...@maven.apache.org > >
