| *** odyssey4me has quit IRC | 00:35 | |
| *** odyssey4me has joined #openstack-requirements | 00:35 | |
| *** hongbin has joined #openstack-requirements | 01:12 | |
| *** hongbin has quit IRC | 03:45 | |
| *** udesale has joined #openstack-requirements | 04:35 | |
| openstackgerrit | garyk proposed openstack/requirements master: Use the correct version for vmware-nsxlib https://review.openstack.org/556175 | 05:22 |
|---|---|---|
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: Updated from generate-constraints https://review.openstack.org/555998 | 06:18 |
| *** toabctl has joined #openstack-requirements | 06:32 | |
| *** andreas_s has joined #openstack-requirements | 06:50 | |
| *** ralonsoh has joined #openstack-requirements | 07:26 | |
| *** markmcclain has quit IRC | 07:40 | |
| *** dhellmann has quit IRC | 07:41 | |
| openstackgerrit | Merged openstack/requirements master: Update G-R for oslo.concurrency>=3.26.0 https://review.openstack.org/555980 | 07:51 |
| *** jpich has joined #openstack-requirements | 07:58 | |
| *** finucannot is now known as stephenfin | 09:03 | |
| *** udesale_ has joined #openstack-requirements | 09:10 | |
| *** udesale has quit IRC | 09:10 | |
| *** udesale_ has quit IRC | 09:15 | |
| *** udesale has joined #openstack-requirements | 09:16 | |
| *** dhellmann has joined #openstack-requirements | 09:31 | |
| *** markmcclain has joined #openstack-requirements | 09:32 | |
| *** jpich has quit IRC | 10:03 | |
| *** jpich has joined #openstack-requirements | 10:08 | |
| *** jpich has joined #openstack-requirements | 10:10 | |
| *** jpich has quit IRC | 10:11 | |
| *** jpich has joined #openstack-requirements | 10:15 | |
| *** jpich has quit IRC | 10:16 | |
| *** jpich has joined #openstack-requirements | 10:17 | |
| *** jpich has quit IRC | 10:21 | |
| *** jpich has joined #openstack-requirements | 10:21 | |
| *** jpich has quit IRC | 10:22 | |
| *** jpich has joined #openstack-requirements | 10:30 | |
| *** jpich has quit IRC | 10:31 | |
| *** andreas_s_ has joined #openstack-requirements | 11:33 | |
| *** andreas_s has quit IRC | 11:35 | |
| openstackgerrit | Merged openstack/requirements master: Bump minimum Django version to 1.11 https://review.openstack.org/556047 | 11:48 |
| *** andreas_s_ has quit IRC | 12:08 | |
| *** andreas_s has joined #openstack-requirements | 12:09 | |
| *** andreas_s has quit IRC | 12:09 | |
| *** andreas_s has joined #openstack-requirements | 12:09 | |
| *** andreas_s_ has joined #openstack-requirements | 12:12 | |
| *** edmondsw has joined #openstack-requirements | 12:13 | |
| *** andreas_s has quit IRC | 12:15 | |
| *** snapiri has quit IRC | 12:16 | |
| *** jpich has joined #openstack-requirements | 12:18 | |
| *** andreas_s has joined #openstack-requirements | 12:34 | |
| *** andreas_s_ has quit IRC | 12:37 | |
| *** odyssey4me has quit IRC | 12:47 | |
| *** odyssey4me has joined #openstack-requirements | 12:47 | |
| *** udesale has quit IRC | 13:53 | |
| *** udesale has joined #openstack-requirements | 13:55 | |
| *** cjloader has joined #openstack-requirements | 14:24 | |
| *** jpich has quit IRC | 14:58 | |
| *** jpich has joined #openstack-requirements | 15:08 | |
| *** jpich has quit IRC | 15:12 | |
| *** jpich has joined #openstack-requirements | 15:13 | |
| *** jpich has quit IRC | 15:14 | |
| *** jpich has joined #openstack-requirements | 15:15 | |
| *** jpich has quit IRC | 15:15 | |
| *** jpich has joined #openstack-requirements | 15:16 | |
| *** jpich has quit IRC | 15:17 | |
| *** jpich has joined #openstack-requirements | 15:19 | |
| *** jpich has quit IRC | 15:19 | |
| *** jpich has joined #openstack-requirements | 15:25 | |
| openstackgerrit | Matthew Thode proposed openstack/requirements master: Updated from generate-constraints https://review.openstack.org/555998 | 15:28 |
| *** jpich has quit IRC | 15:38 | |
| *** andreas_s has quit IRC | 15:39 | |
| *** andreas_s has joined #openstack-requirements | 15:40 | |
| *** jpich has joined #openstack-requirements | 15:48 | |
| *** jpich has quit IRC | 15:48 | |
| *** andreas_s has quit IRC | 15:48 | |
| *** jpich has joined #openstack-requirements | 15:50 | |
| *** jpich has quit IRC | 15:50 | |
| *** AJaeger has joined #openstack-requirements | 15:50 | |
| AJaeger | requirements team, please review https://review.openstack.org/555426 - are you fine removing the requirements sync also for stable releases? | 15:50 |
| AJaeger | prometheanfire: ^ | 15:50 |
| *** jpich has joined #openstack-requirements | 15:51 | |
| *** jpich has quit IRC | 15:52 | |
| *** jpich has joined #openstack-requirements | 15:52 | |
| *** andreas_s has joined #openstack-requirements | 15:54 | |
| *** pabelanger has quit IRC | 15:56 | |
| prometheanfire | not for stable | 15:56 |
| prometheanfire | iirc it was only the plan going forward | 15:56 |
| AJaeger | prometheanfire: then -1 please | 15:59 |
| *** pabelanger has joined #openstack-requirements | 16:00 | |
| *** udesale has quit IRC | 16:17 | |
| *** udesale has joined #openstack-requirements | 16:17 | |
| *** andreas_s has quit IRC | 16:18 | |
| *** andreas_s has joined #openstack-requirements | 16:22 | |
| *** jpich has quit IRC | 16:32 | |
| *** udesale has quit IRC | 16:35 | |
| *** andreas_s has quit IRC | 16:36 | |
| *** andreas_s has joined #openstack-requirements | 16:55 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: update constraint for os-brick to new release 2.4.0 https://review.openstack.org/556590 | 17:02 |
| prometheanfire | dhellmann: dirk tonyb smcginnis - think it's time to retire newton? https://review.openstack.org/555802 | 17:07 |
| *** ralonsoh has quit IRC | 17:08 | |
| *** andreas_s has quit IRC | 17:09 | |
| *** andreas_s has joined #openstack-requirements | 17:14 | |
| smcginnis | prometheanfire: I think it has been retired for most projects, so probably. | 17:41 |
| smcginnis | prometheanfire: tonyb probably has a better handle on where things are on that. | 17:42 |
| smcginnis | But we said our extended maintenance would start with Ocata, so I think it's probably a good time for newton to go away. | 17:42 |
| *** andreas_s has quit IRC | 17:47 | |
| prometheanfire | yep, agreed | 17:49 |
| dhellmann | ++ | 17:49 |
| openstackgerrit | Merged openstack/requirements master: Use the correct version for vmware-nsxlib https://review.openstack.org/556175 | 17:57 |
| *** snapiri has joined #openstack-requirements | 18:58 | |
| *** snapiri has quit IRC | 19:15 | |
| *** EmilienM has quit IRC | 19:57 | |
| *** EmilienM has joined #openstack-requirements | 19:58 | |
| *** edmondsw has quit IRC | 20:21 | |
| openstackgerrit | Merged openstack/requirements master: Updated from generate-constraints https://review.openstack.org/555998 | 20:25 |
| dhellmann | prometheanfire : let me know if/when you have time to talk about https://review.openstack.org/#/c/555426/2 and stable branches | 20:40 |
| prometheanfire | dhellmann: yo, now's good | 20:41 |
| dhellmann | hey | 20:41 |
| prometheanfire | may even catch both tonyb and dirk | 20:41 |
| dhellmann | how often do we change the lower bounds of dependencies on stable branches? | 20:41 |
| prometheanfire | lower bounds, never, but we do add exclusions sometimes | 20:42 |
| * prometheanfire thinks libvirt-python-4.1.0 should be backported to queens/pike | 20:42 | |
| dhellmann | how big of a deal is it to not sync those? because they end up applying to the constraints list either way, right? | 20:43 |
| prometheanfire | it's probably fine, maybe I'm just stuck in my ways | 20:44 |
| prometheanfire | I would like an ack from dirk/tony though | 20:44 |
| dhellmann | oh, yes, we should all agree, definitely | 20:45 |
| dhellmann | and maybe I'm missing an important case | 20:45 |
| dhellmann | it just seems like those should be pretty, well, stable :-) | 20:45 |
| dhellmann | I would be OK with leaving the tools around and just turning off the job | 20:46 |
| prometheanfire | they are | 20:46 |
| dhellmann | that way we could do a sync by hand if we have to | 20:46 |
| prometheanfire | hmm, https://review.openstack.org/556642 didn't show up | 20:46 |
| prometheanfire | maybe I forgot to remove the change-id? | 20:46 |
| prometheanfire | sure | 20:47 |
| dirk | I'll likely be back tomorrow | 20:47 |
| openstackgerrit | Merged openstack/requirements master: Drop ordereddict from g-r https://review.openstack.org/550732 | 20:47 |
| *** edmondsw has joined #openstack-requirements | 20:47 | |
| dirk | I haven't had a chance to be concentrated enough to read reviews that are beyond trivial | 20:48 |
| dhellmann | ok | 20:48 |
| dhellmann | the thing I want to avoid is having a bunch of teams with lower-constraints jobs that pass and bot proposed patches that only change requirements lists and not lower-constraints.txt | 20:48 |
| prometheanfire | that's a good point too | 20:49 |
| dhellmann | in part because I think pip's resolver allows a constraint to be outside of the range (so I need to fix up the patch in the requirements repo) | 20:51 |
| dhellmann | I'll try and do that tomorrow morning | 20:51 |
| prometheanfire | ya, uc overrides ALL | 20:51 |
| dhellmann | so we need the check-requirements job to look at the l-c value and ensure it is in the required range | 20:52 |
| dhellmann | I thought pip did that, so didn't implement that change myself | 20:52 |
| dirk | dhellmann: yeah, that's what's confusing me as well. All the validate checks on the g-r.txt need to be repeated now on the projects repo, right? | 20:53 |
| dhellmann | well, not quite all of them | 20:53 |
| dirk | E.g. only one lower bound, no exclusion below lower bound and so on | 20:53 |
| dhellmann | we're comparing the local requirements against the g-r list | 20:53 |
| dhellmann | so that will take care of most of the checks, right? | 20:53 |
| dirk | Are we comparing that the r. Lower bound is equal to lc? | 20:54 |
| dhellmann | if I say set B has to be a subset of set A, then you can't have things in B that are not in A | 20:54 |
| dhellmann | no, that's the 1 thing I realize now I need to add | 20:54 |
| dirk | And are we checking that all that is in r is also in lc? | 20:54 |
| dhellmann | anything in the local requirements file that is also in the lower constraint list has to have the lower bound match the lower constraint if there is a lower bound | 20:54 |
| dirk | Right, but there isn't a check for that? | 20:55 |
| dhellmann | there is not today, no | 20:55 |
| dhellmann | that's what I'm going to do tomorrow | 20:55 |
| dhellmann | and I can add the check that all r are in lc | 20:55 |
| dirk | Also, the requirements check currently only triggers when certain files are touched | 20:55 |
| dhellmann | that should still be ok, right? | 20:56 |
| dhellmann | I'm not sure all requirements need to be in lower constraints | 20:56 |
| dirk | It needs to cover l-c as well, right? | 20:56 |
| dirk | Imho they should be | 20:56 |
| dhellmann | all of the things in requirements.txt do, but that job tests all of the various requirements files right now | 20:56 |
| dhellmann | the doc/requirements.txt items don't need to be | 20:56 |
| dhellmann | oh, I see what you mean about the check job | 20:57 |
| dirk | Why? | 20:57 |
| dhellmann | yes, I can update it to trigger on the lower-constraints.txt file | 20:57 |
| dhellmann | why are the doc settings tested against the g-r list? | 20:57 |
| dirk | Because we want docs to be buildable with coinstallable deps? | 20:58 |
| dirk | Not sure, wasn't involved in that | 20:58 |
| dhellmann | I guess I can see the benefit of that. | 20:58 |
| dhellmann | the lower-constraints list isn't meant to be a set of co-installable things, though | 20:59 |
| dhellmann | so we don't need to test the doc settings against the l-c list | 20:59 |
| dirk | Also, last question from the trivial kind: what if we want to remove an exclude from g-r? We need to manually update all consumers? | 20:59 |
| dhellmann | under the rules so far, yes, we would have to do that | 20:59 |
| dhellmann | that's the same as what we have today, right? | 21:00 |
| dhellmann | well, except that a bot would do it | 21:00 |
| dhellmann | like I said, we can keep the tools around to automate that, if you want | 21:00 |
| dirk | Yes, which is a big work saver | 21:00 |
| dhellmann | well, ideally we wouldn't need to update all of those projects in the first place | 21:01 |
| dhellmann | we would update g-r and u-c and that would be it | 21:01 |
| dirk | I didn't quite understand the reason for the lc.txt in the projects yet | 21:01 |
| dhellmann | and then to undo it we can just change those lists again | 21:01 |
| dhellmann | the l-c file in the project is to test that the project has accurate lower bounds in its requirements files | 21:01 |
| dirk | That's the other thought I had: what ensures that at the time of a release the git tag being consistent and coinstallable? | 21:02 |
| dhellmann | for now we're only using unit tests, although projects are free to add functional test jobs, too | 21:02 |
| dhellmann | what ensures that today? | 21:02 |
| dirk | Assuming they have mergd the last bot update, the bot update | 21:02 |
| dhellmann | a dsvm gate job, right? | 21:02 |
| dhellmann | no, the bot update does not actually have anything to do with co-installability | 21:03 |
| dhellmann | the u-c list is the co-installability list | 21:03 |
| dhellmann | the constraints completely override any range settings from within the projects | 21:03 |
| dirk | Sure | 21:03 |
| dirk | Bit for users (I heard they exist) only the pip requirements matter | 21:04 |
| dirk | So those need to be consistent with the constraints and coinstallable with a sensible set of other projects at the time of a release of a module | 21:04 |
| dhellmann | we want to make those lists as accurate as possible, but we document that the u-c list is our list of things we actually test with | 21:04 |
| dirk | Currently this is only ensured by freeze periods | 21:05 |
| *** edmondsw has quit IRC | 21:05 | |
| dhellmann | ideally this will mean fewer changes to the requirements files within projects | 21:05 |
| dhellmann | and so we should actually stay coinstallable more of the time | 21:05 |
| dhellmann | we're still going to have a feeze period | 21:05 |
| dirk | Anyway, sorry for speaking up while half being faded out already. I am hopeful to have a look at this when I am back in the office tomorrow | 21:07 |
| dhellmann | yeah, please do. and if you have more detailed questions like this about the process please post on the thread on the mailing list instead of one of the reviews, since it's easier to discuss there | 21:08 |
| dirk | So the lc.txt in the project are not synced across projects | 21:08 |
| dhellmann | I think I understand your concerns, but I think some are about things we don't promise today and some are things that aren't going to change | 21:08 |
| dhellmann | that's right. each project has its own lc.txt file | 21:08 |
| *** edmondsw has joined #openstack-requirements | 21:08 | |
| tonyb | prometheanfire: No, we can't returen newton. We did that with mitaka and broke stuff :( | 21:12 |
| prometheanfire | tonyb: I mean just stop accepting changes, not remove the branch | 21:13 |
| tonyb | prometheanfire: Oh yeah stable/newton pretty much means auto-abandon | 21:14 |
| * tonyb is readyign scrollback for the stop syncing on stable discussion | 21:14 | |
| dhellmann | tonyb : I'm around for a little while longer if you have comments/questions | 21:22 |
| tonyb | dhellmann: I dropped my opinion on the chnage which is "drop the job leave the code" | 21:22 |
| dhellmann | ok | 21:23 |
| dhellmann | wfm | 21:23 |
| tonyb | dhellmann: so that if we need to we can reasonably easily run an update | 21:23 |
| dhellmann | I will abandon the patches to delete the code | 21:23 |
| dhellmann | sure | 21:23 |
| tonyb | dhellmann: it *shouldn't* happen but it probably will :/ | 21:23 |
| tonyb | dhellmann: Thanks. | 21:23 |
| openstackgerrit | Doug Hellmann proposed openstack/requirements master: make the --reqs arg to project-requirements-change.py useful https://review.openstack.org/555403 | 21:25 |
| dhellmann | rebased that fix to keep that one | 21:25 |
| dirk | dhellmann: I appreciate you finally getting a serious dent at working on this, we have not done this for far too long :-/ | 21:26 |
| dhellmann | I'm glad I finally have a little time to spend on it | 21:26 |
| *** edmondsw has quit IRC | 21:26 | |
| dhellmann | I'd like to be done by the first milestone | 21:26 |
| * dhellmann crosses his fingers | 21:26 | |
| dirk | dhellmann: yeah, you're correct to say that we didn't guarantee any of that but the bot updates were usually nagging enough so that people merged them | 21:27 |
| dirk | Which avoided many of the corner cases that we open up to now | 21:27 |
| dhellmann | but the bot updates didn't actually control whether the projects were coinstallable | 21:27 |
| dhellmann | they had no effect at all on that | 21:28 |
| dhellmann | ok; I'm going to drop offline for the evening. tomorrow morning I will work on the changes to the check job that dirk and I discussed | 21:28 |
| dirk | Well, they ensured that a coinstallable set of versions/deps were within declared deps | 21:29 |
| dirk | dhellmann: thanks, have a nice evening. | 21:30 |
| *** cjloader has quit IRC | 22:00 | |
| *** cjloader has joined #openstack-requirements | 22:00 | |
| *** cjloader has quit IRC | 22:04 | |
| *** pabelanger has quit IRC | 23:01 | |
| *** pabelanger has joined #openstack-requirements | 23:01 | |
| *** andreas_s has joined #openstack-requirements | 23:50 | |
| *** andreas_s has quit IRC | 23:54 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!