Triaging: Close as needsinfo
54 views
Skip to first unread message
Shai Berger
May 11, 2013, 4:58:14 PM5/11/13
to django-d...@googlegroups.com
Hi Django devs,
I would like to raise a little concern I ran into lately. When going over some
Oracle-related tickets, I came across ticket #20201[0] . The ticket
description was missing important details, so I commented about them. A few
days later, Aymeric came around, and closed the ticket as "needsinfo",
referencing my comment. To me, this raised a couple of conflicting thoughts:
On one hand -- should I have closed it myself, and saved a core-dev the
trouble? The triaging guide[1] seems very inconclusive about this, advising on
other reasons to close tickets but not on needsinfo (except "tickets which are
really feature requests").
On the other hand -- especially, in the wake of the "Perception of attitude in
tickets" brouhaha -- should we really be closing tickets on needsinfo? In
other communities, I have usually seen "needsinfo" as a ticket state, rather
than a reason for closing; such tickets are then closed later, if enough time
has passed and no further info is received. I realise that, when the person
closing the ticket specifically says "when you have more info, please re-open"
(as Aymeric did in this case), then the process is essentially the same, only
a little more efficient; but I worry that this efficiency may carry a price in, as
Simon put it, "perception of attitude in tickets".
For what it's worth, I, personally, would feel much more at ease setting
tickets as "needsinfo" without closing them -- and if it's a ticket state,
there should be no problem taking such tickets out of the reports developers
use, so they're almost as far out of the way as closed tickets.
Thanks for your consideration,
Shai.
[0] https://code.djangoproject.com/ticket/20201
[1] https://docs.djangoproject.com/en/dev/internals/contributing/triaging-
tickets/
I would like to raise a little concern I ran into lately. When going over some
Oracle-related tickets, I came across ticket #20201[0] . The ticket
description was missing important details, so I commented about them. A few
days later, Aymeric came around, and closed the ticket as "needsinfo",
referencing my comment. To me, this raised a couple of conflicting thoughts:
On one hand -- should I have closed it myself, and saved a core-dev the
trouble? The triaging guide[1] seems very inconclusive about this, advising on
other reasons to close tickets but not on needsinfo (except "tickets which are
really feature requests").
On the other hand -- especially, in the wake of the "Perception of attitude in
tickets" brouhaha -- should we really be closing tickets on needsinfo? In
other communities, I have usually seen "needsinfo" as a ticket state, rather
than a reason for closing; such tickets are then closed later, if enough time
has passed and no further info is received. I realise that, when the person
closing the ticket specifically says "when you have more info, please re-open"
(as Aymeric did in this case), then the process is essentially the same, only
a little more efficient; but I worry that this efficiency may carry a price in, as
Simon put it, "perception of attitude in tickets".
For what it's worth, I, personally, would feel much more at ease setting
tickets as "needsinfo" without closing them -- and if it's a ticket state,
there should be no problem taking such tickets out of the reports developers
use, so they're almost as far out of the way as closed tickets.
Thanks for your consideration,
Shai.
[0] https://code.djangoproject.com/ticket/20201
[1] https://docs.djangoproject.com/en/dev/internals/contributing/triaging-
tickets/
Łukasz Rekucki
May 11, 2013, 5:38:19 PM5/11/13
to django-developers
Hi,
On 11 May 2013 22:58, Shai Berger <sh...@platonix.com> wrote:
In
other communities, I have usually seen "needsinfo" as a ticket state, rather
than a reason for closing; such tickets are then closed later, if enough time
has passed and no further info is received.
To me that's just giving false hope for people viewing that ticket. If the ticket is open, there's a reasonable expectation that there is some action you can take to make it progress, but you most likely you can't as only the reporter has the needed info. For bugs, "needsinfo" is almost identical to "cannot reproduce" (aka "worksforme") and I can't think of a reason to keep such a bug open.
Of course, there are people who will take everything personal. Even if it's getting your ticket closed by a stranger on a public bug tracker.
Łukasz Rekucki
Shai Berger
May 11, 2013, 6:01:34 PM5/11/13
to django-d...@googlegroups.com
On Sunday 12 May 2013, Łukasz Rekucki wrote:
> Hi,
>
> On 11 May 2013 22:58, Shai Berger <sh...@platonix.com> wrote:
> > In
> > other communities, I have usually seen "needsinfo" as a ticket state,
> > rather
> > than a reason for closing; such tickets are then closed later, if enough
> > time
> > has passed and no further info is received.
>
> To me that's just giving false hope for people viewing that ticket. If the
> ticket is open, there's a reasonable expectation that there is some action
> you can take to make it progress, but you most likely you can't as only the
> reporter has the needed info.
I disagree; every person who encounters a problem consistent with the existing
> Hi,
>
> On 11 May 2013 22:58, Shai Berger <sh...@platonix.com> wrote:
> > In
> > other communities, I have usually seen "needsinfo" as a ticket state,
> > rather
> > than a reason for closing; such tickets are then closed later, if enough
> > time
> > has passed and no further info is received.
>
> To me that's just giving false hope for people viewing that ticket. If the
> ticket is open, there's a reasonable expectation that there is some action
> you can take to make it progress, but you most likely you can't as only the
> reporter has the needed info.
description, can provide more info. This is, IMO, the only real (rather than
perceived) difference between closing as "needsinfo" and keeping the ticket in
some sort of open limbo -- if someone else runs into the same problem, I'd
rather they enhance an existing report than start a new one; and nobody looks
at closed tickets.
> For bugs, "needsinfo" is almost identical to
> "cannot reproduce" (aka "worksforme") and I can't think of a reason to keep
> such a bug open.
>
> Of course, there are people who will take everything personal. Even if it's
> getting your ticket closed by a stranger on a public bug tracker.
and the closure seems arbitrary and off-handed, they will be offended. I would
be. Then again, I would actually read the comments on the ticket...
Shai.
Carl Meyer
May 12, 2013, 9:53:47 PM5/12/13
to django-d...@googlegroups.com
On 05/11/2013 06:01 PM, Shai Berger wrote:
> On Sunday 12 May 2013, Łukasz Rekucki wrote:
>> Hi,
>>
>> On 11 May 2013 22:58, Shai Berger <sh...@platonix.com> wrote:
>>> In
>>> other communities, I have usually seen "needsinfo" as a ticket state,
>>> rather
>>> than a reason for closing; such tickets are then closed later, if enough
>>> time
>>> has passed and no further info is received.
>>
>> To me that's just giving false hope for people viewing that ticket. If the
>> ticket is open, there's a reasonable expectation that there is some action
>> you can take to make it progress, but you most likely you can't as only the
>> reporter has the needed info.
>
> I disagree; every person who encounters a problem consistent with the existing
> description, can provide more info. This is, IMO, the only real (rather than
> perceived) difference between closing as "needsinfo" and keeping the ticket in
> some sort of open limbo -- if someone else runs into the same problem, I'd
> rather they enhance an existing report than start a new one; and nobody looks
> at closed tickets.
It seems this is the only argument for not having "needsinfo" as a
> On Sunday 12 May 2013, Łukasz Rekucki wrote:
>> Hi,
>>
>> On 11 May 2013 22:58, Shai Berger <sh...@platonix.com> wrote:
>>> In
>>> other communities, I have usually seen "needsinfo" as a ticket state,
>>> rather
>>> than a reason for closing; such tickets are then closed later, if enough
>>> time
>>> has passed and no further info is received.
>>
>> To me that's just giving false hope for people viewing that ticket. If the
>> ticket is open, there's a reasonable expectation that there is some action
>> you can take to make it progress, but you most likely you can't as only the
>> reporter has the needed info.
>
> I disagree; every person who encounters a problem consistent with the existing
> description, can provide more info. This is, IMO, the only real (rather than
> perceived) difference between closing as "needsinfo" and keeping the ticket in
> some sort of open limbo -- if someone else runs into the same problem, I'd
> rather they enhance an existing report than start a new one; and nobody looks
> at closed tickets.
closed state, and it rests on a premise ("nobody looks at closed
tickets") that is contradicted by the significant activity we regularly
see on closed tickets. Most people find tickets via Google or links on
e.g. Stack Overflow, and these do not discriminate between closed and
open tickets. I also have not observed a pattern of needsinfo tickets
collecting more dupes than any other type of ticket.
>> Of course, there are people who will take everything personal. Even if it's
>> getting your ticket closed by a stranger on a public bug tracker.
>
> When people have already taken the trouble to file a ticket, if it is closed
> and the closure seems arbitrary and off-handed, they will be offended. I would
> be. Then again, I would actually read the comments on the ticket...
an explicit "please reopen if you can provide more info." This seems
good enough to me; I haven't observed a problem of people responding
poorly to a "needsinfo" closing (the issues are always around "wontfix"
closings). Making the change you are suggesting means introducing a new
tracker-gardening task, "closing old needsinfo tickets." Needsinfo was
added as a "closed" state precisely in order to get rid of the need for
this.
(And FWIW, community triagers like yourself should feel entirely free to
close tickets needsinfo. I think the only reason that isn't clearer in
the contributing guide is that the needsinfo status was added relatively
recently.)
Carl
Aymeric Augustin
May 13, 2013, 4:53:56 PM5/13/13
to django-d...@googlegroups.com
Hi Shai,
On 11 mai 2013, at 22:58, Shai Berger <sh...@platonix.com> wrote:
> On one hand -- should I have closed it myself, and saved a core-dev the
> trouble? The triaging guide[1] seems very inconclusive about this, advising on
> other reasons to close tickets but not on needsinfo (except "tickets which are
> really feature requests").
Yes, this is the best course of action. I'll clarify the contributing guide.
There are several reasons for this:
- it increases the signal-noise ratio in open tickets,
- often the reporter never responds because he moved on and no longer cares,
- we have to keep some balance between the efforts required from people who
report bugs and those who attempt to reproduce them.
For this ticket, if the reporter were paying my consulting rate — and that's
not far fetched, he can afford Oracle licenses after all — he would certainly
provide a more thorough description of his problem. Giving away our time
doesn't make it less respectable or valuable.
> On the other hand -- especially, in the wake of the "Perception of attitude in
> tickets" brouhaha -- should we really be closing tickets on needsinfo? In
> other communities, I have usually seen "needsinfo" as a ticket state, rather
> than a reason for closing; such tickets are then closed later, if enough time
> has passed and no further info is received. I realise that, when the person
> closing the ticket specifically says "when you have more info, please re-open"
> (as Aymeric did in this case), then the process is essentially the same, only
> a little more efficient; but I worry that this efficiency may carry a price in, as
> Simon put it, "perception of attitude in tickets".
When a ticket is closed a "needsinfo", it also carries the message that the
ticket wasn't reported properly, and thus wasted a bit of everyone's time.
Closing the ticket reflects this, for better or worse.
> For what it's worth, I, personally, would feel much more at ease setting
> tickets as "needsinfo" without closing them -- and if it's a ticket state,
> there should be no problem taking such tickets out of the reports developers
> use, so they're almost as far out of the way as closed tickets.
I don't think it's appropriate to ignore open tickets. If we're going to ignore
them, let's close them.
Plus, everything Carl said, and which I removed from my draft :)
--
Aymeric.
On 11 mai 2013, at 22:58, Shai Berger <sh...@platonix.com> wrote:
> On one hand -- should I have closed it myself, and saved a core-dev the
> trouble? The triaging guide[1] seems very inconclusive about this, advising on
> other reasons to close tickets but not on needsinfo (except "tickets which are
> really feature requests").
There are several reasons for this:
- it increases the signal-noise ratio in open tickets,
- often the reporter never responds because he moved on and no longer cares,
- we have to keep some balance between the efforts required from people who
report bugs and those who attempt to reproduce them.
For this ticket, if the reporter were paying my consulting rate — and that's
not far fetched, he can afford Oracle licenses after all — he would certainly
provide a more thorough description of his problem. Giving away our time
doesn't make it less respectable or valuable.
> On the other hand -- especially, in the wake of the "Perception of attitude in
> tickets" brouhaha -- should we really be closing tickets on needsinfo? In
> other communities, I have usually seen "needsinfo" as a ticket state, rather
> than a reason for closing; such tickets are then closed later, if enough time
> has passed and no further info is received. I realise that, when the person
> closing the ticket specifically says "when you have more info, please re-open"
> (as Aymeric did in this case), then the process is essentially the same, only
> a little more efficient; but I worry that this efficiency may carry a price in, as
> Simon put it, "perception of attitude in tickets".
ticket wasn't reported properly, and thus wasted a bit of everyone's time.
Closing the ticket reflects this, for better or worse.
> For what it's worth, I, personally, would feel much more at ease setting
> tickets as "needsinfo" without closing them -- and if it's a ticket state,
> there should be no problem taking such tickets out of the reports developers
> use, so they're almost as far out of the way as closed tickets.
them, let's close them.
Plus, everything Carl said, and which I removed from my draft :)
--
Aymeric.
Reply all
Reply to author
Forward
0 new messages