Stalled tickets

136 views
Skip to first unread message

David Vaz

unread,
Sep 26, 2019, 7:45:52 PM9/26/19
to Django developers (Contributions to Django itself)
Hi,

I am playing around at the DjangoCon US 2019 Sprints and while trying to do my share one thing stands out in the open tickets is: Some are very old and others have not been touched in a while. 

I did a simple analysis of accepted and open tickets based on last modified time (kind of a live status).

The numbers might change over time, but I also linked them: 

Modified at any time All: 1264

Modified in last 5 years: 1039

Modified in last 4 years:  887

Modified in last 3 years:  728

Modified in last 2 years:  494

Modified in last year:     329


So if we would decide to close stalled tickets after some inactivity period we could massively reduce the opened tickets list. Imagine if we close any ticket no modified in the last 5 years we would reduce by 20% the active ticket list. If we decide to be more aggressive, say 3 years we can cut by half the active tickets list. 


I also believe that apparently stalled tickets might be important. This auto-close approach would also trigger a live prof, any automatically closed ticket could be reopened if relevant.


Let us focus the efforts on the really active ones.


Anyway, these are just my thoughts. 


For those of you who do not know me, I am organizing DjangoCon Europe 2020 in Porto Portugal and you are all invited, official details are about to be released.

ludovic coues

unread,
Sep 27, 2019, 2:49:11 AM9/27/19
to django-d...@googlegroups.com
I have seen other open source project handling that with a comment saying the ticket will be closed in a short time. I assume closing with a comment it's fine to reopen if it's still relevant would be fine. Maybe also tagging the tickets with a label "closed as stalled" ?

--
You received this message because you are subscribed to the Google Groups "Django developers (Contributions to Django itself)" group.
To unsubscribe from this group and stop receiving emails from it, send an email to django-develop...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/django-developers/b3b855e5-7fd6-4b26-b365-75ff687c4b4c%40googlegroups.com.

Claude Paroz

unread,
Sep 27, 2019, 3:08:53 AM9/27/19
to Django developers (Contributions to Django itself)
Sorry but I would *strongly* oppose any idea of closing a ticket because noone commented in the last x years. Most of those tickets are still valid issues, but they remain open because noone dedicated the appropriate time to solve them, sometimes because the problem is a corner case, sometimes because it would take considerable time to prepare a reviewable patch.

I don't know what it would bring to the project to close those stalled issues. I still hope that more people would dedicate some of their time (free or paid) to "adopt" one of those tickets and make the effort to lead them until resolution.

You've been invited :-)

Claude

Adam Johnson

unread,
Sep 27, 2019, 5:33:04 AM9/27/19
to django-d...@googlegroups.com
I'm also not sure of the value of closing stalled tickets. Closing them makes them less visible, and as Claude says, they're still valid. Often I find one that's been open years has "come up again" on the mailing list or client work.

The fellows are the ones who work the most with ticketing system, I've never heard any past or present fellow say that it would be a good idea to close open, valid tickets. They're also very good at closing tickets they believe are invalid.

--
You received this message because you are subscribed to the Google Groups "Django developers (Contributions to Django itself)" group.
To unsubscribe from this group and stop receiving emails from it, send an email to django-develop...@googlegroups.com.


--
Adam

Elena Williams

unread,
Sep 27, 2019, 6:51:39 AM9/27/19
to django-d...@googlegroups.com
Interestingly there's an example of a long dead ticket rising 7-8 days ago: https://code.djangoproject.com/ticket/14218

I believe this is an interesting reference for this conversation though not sure if any side of the discussion is helped.

---
Elena Williams
Github: elena


Mariusz Felisiak

unread,
Sep 27, 2019, 7:17:32 AM9/27/19
to Django developers (Contributions to Django itself)
I'm also strongly against closing an "idle" tickets. Sometimes tickets are solved after many years (even 9, 10, or more). Age doesn't make them less valid. In "open tickets scale" 4-5 years is not a long time. Closing tickets just because they're old or because we would like to get a better statistics is not a good solution. Believe me, I'm committed to decrease the number of accepted tickets but without harming Django. It is a long process, we were able to reduce the no. of open tickets by ~70 in the last six months what IMO is a great achievement.

Best,
Mariusz

Virgílio Santos

unread,
Sep 27, 2019, 8:55:00 AM9/27/19
to django-d...@googlegroups.com
Hello,

I'm looking for someone to walk me through a fast path to getting active on the community and start solving tickets! 

I know there is a lot of docs, but it will be extra time of my working day, so it will be very nice to be somehow mentored through delivering than reading lots of docs!

That is it, I think it would be better a great effort on mentoring people through delivering focused on old tickets than closing them and cut off their visibility

Virgilio


--

Carlton Gibson

unread,
Sep 27, 2019, 10:53:42 AM9/27/19
to Django developers (Contributions to Django itself)
Hi David. 

First off, well done for getting your first PR merged today! 🎉

Then, thanks for raising this; it's worth talking about at least. 

I too would like to bring the ticket count down, but I would like to do it by fixing them. 🙂

I think there are two approaches here. Certainly on DRF, Tom Christie implemented an approach of closing tickets that weren't really active. 

I've followed similar on Django-Filter, say. 

Historically, that's not what we've gone for on Django. Rather, we've seen no harm in keeping valid tickets open. 
(The filters you link to already let you cut down the number of tickets you see, if that's the goal.) 

For **many** of the tickets, they're not necessarily super hard, they just need time and love. 
So new contributions like yours are the secret. 

It may be a harder job to get the number down this way, but I'm optimistic we can do it. 
Plus Mariusz and I very much enjoy merging fixes to really old tickets. It's somewhat satisfying. 
(I don't guess anyone else notices, but... 🙂) 

I hope that makes sense. 

And thanks for your super contributions this week. Welcome aboard! 🏅

Kind Regards,

Carlton

Peter van der Does

unread,
Sep 27, 2019, 12:25:40 PM9/27/19
to django-d...@googlegroups.com

I agree with most stating that you can't just close tickets, but I do recommend that tickets need to be checked upon and acted on when:

  • A new minor/major version is released. If the ticket is also related to the newer version include this version into the ticket so people know it's still an issue for that version as well.
  • The version related to the ticket reaches end of line, the ticket should be closed.


I believe keeping track of which versions are related to the ticket means adding another field to the ticket, as currently "version" indicates for which version the ticket was submitted.


On 9/26/19 7:42 PM, David Vaz wrote:
Hi,

I am playing around at the DjangoCon US 2019 Sprints and while trying to do my share one thing stands out in the open tickets is: Some are very old and others have not been touched in a while. 

I did a simple analysis of accepted and open tickets based on last modified time (kind of a live status).

The numbers might change over time, but I also linked them: 

Modified at any time All: 1264

Modified in last 5 years: 1039

Modified in last 4 years:  887

Modified in last 3 years:  728

Modified in last 2 years:  494

Modified in last year:     329


So if we would decide to close stalled tickets after some inactivity period we could massively reduce the opened tickets list. Imagine if we close any ticket no modified in the last 5 years we would reduce by 20% the active ticket list. If we decide to be more aggressive, say 3 years we can cut by half the active tickets list. 



--
Peter van der Does
o: 
410-584-2500
ONeil Interactive, Inc 

Tobias Kunze

unread,
Sep 27, 2019, 2:53:59 PM9/27/19
to django-d...@googlegroups.com
Hi David,

first off: thank you for voicing your opinion and starting this discussion.
Project governance decisions like this are often implicit or dictated by
tradition, so it's worth revisiting them occasionally!

On 19-09-26 16:42:25, David Vaz wrote:
>So if we would decide to close stalled tickets after some inactivity period
>we could massively reduce the opened tickets list.

I'm strongly opposed to this move. In all Open Source projects I have ever
seen with this pattern, it was a frustrating experience for all involved:

As a user I'd report a bug, the developers would confirm it, and if they
didn't get round to fixing it, I'd receive a notification that the issue was
about to be closed unless I took action. So, every two months or so, I'd drop
a "the issue still persists" comment (annoying, really!) – or I'd be so
frustrated that I'd just not respond and see the issue closed.

As a user who found a bug, I'd have to go through a LOT of closed-but-valid
tickets to see if mine was a duplicate – these systems tend to have a lot more
duplicates, naturally, since many people don't go through this touble.

And as a developer, you get a notification every time somebody comments "issue
still persists" (often in increasingly clipped or frustrated messages), which
is not exactly what you want to wake up to either.

>Let us focus the efforts on the really active ones.

Or: Let us focus the efforts on the really valid ones. And if all the open
ones are valid, I see no advantage to closing them.

Best regards,
Tobias
signature.asc

Adam Johnson

unread,
Sep 27, 2019, 3:49:59 PM9/27/19
to Tobias Kunze, django-d...@googlegroups.com
Hi Peter,

I do recommend that tickets need to be checked upon and acted on when:

A new minor/major version is released. If the ticket is also related to the newer version include this version into the ticket so people know it's still an issue for that version as well.
The version related to the ticket reaches end of line, the ticket should be closed.

I'd estimate with the number of open tickets we have, that'd be about the same amount of work as the fellows already put into each version. It'd be nice to do for sure but there just aren't the resources.

Also I think the majority of tickets filed against e.g. 1.8 are still valid today. Bug fixes occasionally fix other bugs, but new features rarely supplant other feature requests.

Thanks,

Adam

--
You received this message because you are subscribed to the Google Groups "Django developers  (Contributions to Django itself)" group.
To unsubscribe from this group and stop receiving emails from it, send an email to django-develop...@googlegroups.com.


--
Adam
Reply all
Reply to author
Forward
0 new messages