[Twisted-Python] Handling upstream patches
Hi, I have received notification for the following LP/Ubuntu issues https://bugs.launchpad.net/ubuntu/+source/twisted/+bug/1505748 https://bugs.launchpad.net/maas/+bug/1504971 ------ I think that they are related to http://twistedmatrix.com/trac/ticket/7650 ------- As a Twisted committer do I need to care about LP / Ubuntu upstream patches? Does Twisted has a follow-up process so that upstream patches are applied in trunk? Thanks! -- Adi Roiban
On 16 October 2015 at 12:07, Adi Roiban
As a Twisted committer do I need to care about LP / Ubuntu upstream patches?
Does Twisted has a follow-up process so that upstream patches are applied in trunk?
I think we (the MAAS team) should propose our patch for inclusion into Twisted. https://code.launchpad.net/~jtv/maas/bug-1372767/+merge/235616 is the original fix within MAAS, which monkey-patches twisted.web.client.URI/_URI. This has the benefit of having tests which the patch that made it into Ubuntu lacks. I'll put together a patch that can be applied to Twisted proper (i.e. the patch that has gone into Ubuntu, plus tests) and propose it. Gavin.
On Oct 16, 2015, at 6:23 AM, Gavin Panella
wrote: On 16 October 2015 at 12:07, Adi Roiban
wrote: As a Twisted committer do I need to care about LP / Ubuntu upstream patches?
Does Twisted has a follow-up process so that upstream patches are applied in trunk?
I think we (the MAAS team) should propose our patch for inclusion into Twisted.
https://code.launchpad.net/~jtv/maas/bug-1372767/+merge/235616 is the original fix within MAAS, which monkey-patches twisted.web.client.URI/_URI. This has the benefit of having tests which the patch that made it into Ubuntu lacks.
I'll put together a patch that can be applied to Twisted proper (i.e. the patch that has gone into Ubuntu, plus tests) and propose it.
Generally speaking, we should wait for contributors to submit their patches to Twisted through the normal process and not go grabbing stuff from downstream. Canonical's IP policy is weird, and it is not clear to me that a patch necessarily has to be MIT licensed to be accepted into Ubuntu's Twisted, since Ubuntu itself is a commercial work. -glyph
On 16 October 2015 at 17:50, Glyph Lefkowitz
Canonical's IP policy is weird, and it is not clear to me that a patch necessarily has to be MIT licensed to be accepted into Ubuntu's Twisted, since Ubuntu itself is a commercial work.
Instinctively I would assume that Ubuntu would reject patches that would also be rejected by upstream on licensing grounds. It's in Ubuntu's and Canonical's interest to contribute upstream, and meddling with licenses would hamper that. The "weird" IP policy thing might be about things for which Canonical is the upstream, but I honestly don't know much about that and definitely cannot speak authoritatively. To confirm, LaMont -- who submitted the patch under question into Ubuntu -- did ask our team's management, and the answer was that Canonical has no interest in keeping that patch from going upstream, under whatever license applies to Twisted. Gavin.
On Oct 18, 2015, at 6:10 AM, Gavin Panella
wrote: On 16 October 2015 at 17:50, Glyph Lefkowitz
wrote: [...] Canonical's IP policy is weird, and it is not clear to me that a patch necessarily has to be MIT licensed to be accepted into Ubuntu's Twisted, since Ubuntu itself is a commercial work.
Instinctively I would assume that Ubuntu would reject patches that would also be rejected by upstream on licensing grounds. It's in Ubuntu's and Canonical's interest to contribute upstream, and meddling with licenses would hamper that. The "weird" IP policy thing might be about things for which Canonical is the upstream, but I honestly don't know much about that and definitely cannot speak authoritatively.
It is not because I believe that Canonical would deliberately do something bizarre like making their version of Twisted actually be GPL3, but rather, that they make no guarantees to this effect (nor should they have to, the patches are in their system, not ours!).
To confirm, LaMont -- who submitted the patch under question into Ubuntu -- did ask our team's management, and the answer was that Canonical has no interest in keeping that patch from going upstream, under whatever license applies to Twisted.
I'd still prefer that we just accept patches consistently through one mechanism. Forgetting about the license question for a second, although today the review queue happens to be empty (!!!), a patch needs a champion willing to respond to review feedback almost as much as it needs to be written in the first place. If the author doesn't have time to submit it, then it's unlikely they'll have time to respond to feedback and complete the process, which just makes more detritus in our issue tracker. -glyph
On 18 October 2015 at 22:47, Glyph Lefkowitz
I'd still prefer that we just accept patches consistently through one mechanism. Forgetting about the license question for a second, although today the review queue happens to be empty (!!!), a patch needs a champion willing to respond to review feedback almost as much as it needs to be written in the first place. If the author doesn't have time to submit it, then it's unlikely they'll have time to respond to feedback and complete the process, which just makes more detritus in our issue tracker.
I fully agree. We should have pushed this fix into Twisted a long time ago. In the release-date-imminent maelstrom we unfortunately forgot about it at the time, and only noticed it again while porting MAAS to Python 3. I have a patch for ticket #7650 ready which I'll put up for review when Trac decides that it's not full of spam. Gavin.
On Oct 19, 2015, at 3:22 AM, Gavin Panella
wrote: On 18 October 2015 at 22:47, Glyph Lefkowitz
wrote: [...] I'd still prefer that we just accept patches consistently through one mechanism. Forgetting about the license question for a second, although today the review queue happens to be empty (!!!), a patch needs a champion willing to respond to review feedback almost as much as it needs to be written in the first place. If the author doesn't have time to submit it, then it's unlikely they'll have time to respond to feedback and complete the process, which just makes more detritus in our issue tracker.
I fully agree. We should have pushed this fix into Twisted a long time ago. In the release-date-imminent maelstrom we unfortunately forgot about it at the time, and only noticed it again while porting MAAS to Python 3.
I have a patch for ticket #7650 ready which I'll put up for review when Trac decides that it's not full of spam.
The spam filter has been re-set, and I don't see any items in the monitoring queue. Did you try to re-submit it recently and get denied? -glyph
On 20 October 2015 at 00:20, Glyph Lefkowitz
The spam filter has been re-set, and I don't see any items in the monitoring queue. Did you try to re-submit it recently and get denied?
I was denied the once and I didn't immediately retry. I resubmitted the patch later in the day and it was accepted. I assumed that a kind person had trained the filter in the meantime. Gavin.
participants (3)
-
Adi Roiban
-
Gavin Panella
-
Glyph Lefkowitz