| *** dangtrinhnt has quit IRC | 00:20 | |
| *** dangtrinhnt has joined #openstack-tc | 00:36 | |
| *** jamesmcarthur has quit IRC | 00:49 | |
| *** jamesmcarthur has joined #openstack-tc | 00:52 | |
| *** tbarron has joined #openstack-tc | 01:04 | |
| *** jamesmcarthur has quit IRC | 01:24 | |
| *** jamesmcarthur has joined #openstack-tc | 01:45 | |
| *** jamesmcarthur has quit IRC | 02:07 | |
| *** jamesmcarthur has joined #openstack-tc | 02:20 | |
| *** jamesmcarthur has quit IRC | 02:25 | |
| *** jamesmcarthur has joined #openstack-tc | 02:26 | |
| *** jamesmcarthur has quit IRC | 02:31 | |
| *** jamesmcarthur has joined #openstack-tc | 02:46 | |
| *** jamesmcarthur has quit IRC | 03:06 | |
| *** jamesmcarthur_ has joined #openstack-tc | 03:06 | |
| *** jamesmcarthur_ has quit IRC | 03:26 | |
| *** gagehugo has quit IRC | 03:34 | |
| *** jamesmcarthur has joined #openstack-tc | 03:34 | |
| *** gagehugo has joined #openstack-tc | 03:36 | |
| *** jamesmcarthur has quit IRC | 03:52 | |
| *** jamesmcarthur has joined #openstack-tc | 04:00 | |
| *** jamesmcarthur has quit IRC | 04:04 | |
| *** jamesmcarthur has joined #openstack-tc | 04:15 | |
| *** jamesmcarthur has quit IRC | 04:20 | |
| *** jamesmcarthur has joined #openstack-tc | 04:22 | |
| *** jamesmcarthur has quit IRC | 04:51 | |
| *** jamesmcarthur has joined #openstack-tc | 04:58 | |
| *** jamesmcarthur has quit IRC | 05:11 | |
| *** jamesmcarthur has joined #openstack-tc | 05:14 | |
| *** jamesmcarthur has quit IRC | 05:33 | |
| *** jamesmcarthur has joined #openstack-tc | 05:35 | |
| *** jamesmcarthur has quit IRC | 05:52 | |
| *** e0ne has joined #openstack-tc | 06:00 | |
| *** jamesmcarthur has joined #openstack-tc | 06:04 | |
| *** jamesmcarthur has quit IRC | 06:09 | |
| *** jamesmcarthur has joined #openstack-tc | 06:13 | |
| *** jamesmcarthur has quit IRC | 06:17 | |
| *** e0ne has quit IRC | 06:28 | |
| *** jamesmcarthur has joined #openstack-tc | 06:40 | |
| *** jamesmcarthur has quit IRC | 06:45 | |
| *** Bhujay has joined #openstack-tc | 06:47 | |
| *** Bhujay has quit IRC | 06:49 | |
| *** mtreinish has quit IRC | 06:55 | |
| *** mtreinish has joined #openstack-tc | 06:58 | |
| *** leakypipes has quit IRC | 07:10 | |
| *** leakypipes has joined #openstack-tc | 07:11 | |
| *** jamesmcarthur has joined #openstack-tc | 07:12 | |
| *** jamesmcarthur has quit IRC | 07:16 | |
| *** evrardjp_ is now known as evrardjp | 07:21 | |
| *** jamesmcarthur has joined #openstack-tc | 07:36 | |
| *** jamesmcarthur has quit IRC | 07:41 | |
| *** tosky has joined #openstack-tc | 07:45 | |
| *** jamesmcarthur has joined #openstack-tc | 07:54 | |
| *** jamesmcarthur has quit IRC | 07:58 | |
| *** csatari has joined #openstack-tc | 08:01 | |
| *** jamesmcarthur has joined #openstack-tc | 08:04 | |
| *** jpich has joined #openstack-tc | 08:06 | |
| *** jamesmcarthur has quit IRC | 08:08 | |
| *** jamesmcarthur has joined #openstack-tc | 08:27 | |
| *** jamesmcarthur has quit IRC | 08:32 | |
| *** Bhujay has joined #openstack-tc | 08:36 | |
| *** Bhujay has quit IRC | 08:37 | |
| *** jamesmcarthur has joined #openstack-tc | 08:37 | |
| *** Bhujay has joined #openstack-tc | 08:37 | |
| *** jamesmcarthur has quit IRC | 08:41 | |
| *** jamesmcarthur has joined #openstack-tc | 09:14 | |
| openstackgerrit | Hervé Beraud proposed openstack/project-team-guide master: Using pip as a python module https://review.openstack.org/610465 | 09:31 |
|---|---|---|
| *** Bhujay has quit IRC | 09:47 | |
| *** Bhujay has joined #openstack-tc | 09:54 | |
| *** Bhujay has quit IRC | 09:55 | |
| *** Bhujay has joined #openstack-tc | 09:56 | |
| *** leakypipes is now known as jaypipes | 10:26 | |
| *** dtantsur is now known as dtantsur|brb | 11:30 | |
| *** jamesmcarthur has quit IRC | 12:11 | |
| *** jamesmcarthur has joined #openstack-tc | 12:11 | |
| *** jamesmcarthur has quit IRC | 12:30 | |
| *** jroll has quit IRC | 12:38 | |
| *** jroll has joined #openstack-tc | 12:39 | |
| *** EmilienM has quit IRC | 12:46 | |
| *** EmilienM has joined #openstack-tc | 12:47 | |
| *** jamesmcarthur has joined #openstack-tc | 12:54 | |
| *** jaypipes has quit IRC | 12:55 | |
| *** dtantsur|brb is now known as dtantsur | 12:59 | |
| *** mriedem has joined #openstack-tc | 13:19 | |
| *** lbragstad has joined #openstack-tc | 13:20 | |
| *** SteelyDan is now known as dansmith | 13:28 | |
| *** cdent has joined #openstack-tc | 13:56 | |
| *** Bhujay has quit IRC | 14:06 | |
| *** annabelleB has joined #openstack-tc | 14:20 | |
| mriedem | mnaser: so people are having problems with the upgrade-checkers stuff huh? | 14:23 |
| mriedem | i wonder what the intersection is of people that have questions and people that have actually asked me questions | 14:24 |
| * smcginnis imagines two separate circles on the venn diagram | 14:26 | |
| * TheJulia imagines a third circle where people haven't even gotten started on it yet | 14:28 | |
| * TheJulia goes and stands in that third circle | 14:28 | |
| lbragstad | i think we *just* added oslo-upgradecheck to g-r | 14:29 |
| * cdent gives TheJulia a cake | 14:31 | |
| mriedem | lbragstad: we did | 14:31 |
| TheJulia | cdent: oooohhh ahhhhh | 14:31 |
| mriedem | https://review.openstack.org/#/c/610061/ | 14:31 |
| lbragstad | yeah - nice... i just recheck our implementation, which should be in the clear now | 14:32 |
| smcginnis | I was waiting on that for the cinder work that I just submitted this morning. | 14:32 |
| mriedem | so was slaweq for neutron | 14:32 |
| lbragstad | https://review.openstack.org/#/c/608785/ | 14:32 |
| mriedem | i'm just trying to gauge how concerned i should be | 14:32 |
| mriedem | like, do i need to spend a day making educational slides and a video or what | 14:32 |
| smcginnis | https://review.openstack.org/#/c/610581/ | 14:33 |
| smcginnis | I think now that there are a few examples, that might be a good base for others to take a look at and copy. | 14:33 |
| lbragstad | true | 14:33 |
| smcginnis | For the most part, the base framework is pretty generic. | 14:33 |
| lbragstad | also - i'd be curious to hear who needs the help | 14:33 |
| smcginnis | The more difficult part is deciding what and how to check things. | 14:33 |
| lbragstad | then i can carve out a few hours to give them a hand | 14:34 |
| smcginnis | mriedem: Maybe offer to have an "office hour" in #openstack-dev for this? | 14:34 |
| TheJulia | ++ | 14:34 |
| mriedem | sure | 14:36 |
| lbragstad | if you schedule one - i'll attend and take a stab at proprosing patches that add the scaffolding to projects | 14:37 |
| evrardjp | mriedem: blog post? | 14:38 |
| mriedem | there is already docs, | 14:39 |
| mriedem | and a story, | 14:39 |
| mriedem | so i don't see how a blog post with the same words is going to clarify any confusion | 14:39 |
| evrardjp | it's on one place. | 14:39 |
| evrardjp | you can explain that differently. | 14:40 |
| mriedem | the story links to the docs :) | 14:40 |
| mriedem | if anything i think it needs to be a forum if people are super confused | 14:40 |
| mriedem | not The Forum (tm) mind you | 14:40 |
| lbragstad | what about a mini sprint? | 14:41 |
| mriedem | it's been a couple of years since i've been agile | 14:42 |
| evrardjp | I should not have laughed there. | 14:43 |
| mriedem | lbragstad: as in, "everyone that is working on these, do it the first week of november?" or something? | 14:43 |
| mriedem | we're getting close enough to the forum and with enough projects not started, i wonder if there is room for a slot about this for Q&A | 14:43 |
| smcginnis | I wonder how difficult it would be to script up the scaffolding addition. That seemed to work well for Doug with the python3-first work. | 14:48 |
| lbragstad | ++ | 14:49 |
| lbragstad | yeah - essentially an hour dedicated to getting the scaffolding done and reviewing those patches | 14:49 |
| lbragstad | review, respin, and merge | 14:49 |
| lbragstad | then, it might be easier to have conversations about what qualifies as an upgrade check if people have that step done | 14:52 |
| * lbragstad shrugs | 14:52 | |
| zaneb | mriedem: at the risk of misinterpreting mnaser, I'm not sure that there is a problem with implementing the upgrade checker goal per se. but a concern that I and others expressed at the PTG is that multiple goals have been approved before we had turn-key instructions for implementing them | 14:55 |
| zaneb | in the case of WSGI the instructions changed after some people had implemented it. perhaps that case was just bad luck | 14:56 |
| zaneb | but the question for me is how can we make sure that stuff is in place and locked down before we ask the whole community to do it in future | 14:56 |
| cdent | are we really expecting these things to be cookie cutters with recipes? Or is the spirit of the goal more important? | 14:57 |
| zaneb | it depends, I think | 14:58 |
| mriedem | locking down what 50+ different projects need to do with differeing levels of activity and maturity for different goals is going to be hard to lay out | 14:58 |
| mriedem | cdent: i certainly don't ^ | 14:58 |
| lbragstad | the variability in implementations across openstack is going to make that turn-key implementation tough | 14:58 |
| zaneb | there's no cookie-cutter recipe for mox removal, to cite mriedem's favourite example ;) | 14:58 |
| cdent | mriedem: I suspect we agree on this | 14:58 |
| evrardjp | cdent: openstack is not a place for cookie-cuttering many things I am afraid | 14:59 |
| evrardjp | :) | 14:59 |
| mriedem | unless the goal is so dead simple like mox removal, but then it's not really a worthy goal either | 14:59 |
| mriedem | zaneb: you beat me to it :) | 14:59 |
| evrardjp | :) | 14:59 |
| mriedem | so who is the glance liaison from the tc? because i haven't had a response to http://lists.openstack.org/pipermail/openstack-dev/2018-September/135025.html yet | 14:59 |
| mriedem | i'm trying to help projects by going over their release notes to identify check-worthy things and those are documented per project task in the story | 15:00 |
| lbragstad | that would be me | 15:00 |
| zaneb | mriedem: liaisons are being reassigned as we speak after the TC election | 15:00 |
| lbragstad | i just put my name down a few days ago actually | 15:00 |
| mriedem | but i need help from the maintainers of those projects to tell me i'm way off the mark | 15:00 |
| lbragstad | https://wiki.openstack.org/wiki/OpenStack_health_tracker | 15:00 |
| *** Bhujay has joined #openstack-tc | 15:02 | |
| * dims catches up | 15:07 | |
| *** Bhujay has quit IRC | 15:14 | |
| lbragstad | mriedem rosmaita is going to reply to your email | 15:23 |
| *** Bhujay has joined #openstack-tc | 15:32 | |
| *** openstackgerrit has quit IRC | 15:47 | |
| *** openstackgerrit has joined #openstack-tc | 15:48 | |
| mriedem | lbragstad: thanks | 15:53 |
| *** e0ne has joined #openstack-tc | 16:04 | |
| *** e0ne has quit IRC | 16:21 | |
| *** e0ne has joined #openstack-tc | 16:27 | |
| *** e0ne has quit IRC | 16:28 | |
| *** annabelleB has quit IRC | 16:31 | |
| *** jamesmcarthur has quit IRC | 16:39 | |
| *** annabelleB has joined #openstack-tc | 16:39 | |
| *** jpich has quit IRC | 16:44 | |
| *** Bhujay has quit IRC | 16:46 | |
| cdent | zaneb: is there pending state on the tech vision statement? | 17:09 |
| zaneb | cdent: I have a couple of additions ready to go up. I'm also working on emails to every team to explain what the vision means to them. I'm expecting to encounter some more gaps in the process of writing them | 17:11 |
| zaneb | (additions so far are the GUI goal + Jay's section on regions) | 17:11 |
| cdent | zaneb: cool, just wanted to check in as i was away and sort of out of the loop | 17:15 |
| *** dangtrinhnt has quit IRC | 17:19 | |
| *** dangtrinhnt has joined #openstack-tc | 17:19 | |
| TheJulia | ttx: (or any other foundation folks that are aware): When can we expect the summit schedule to be updated with forum sessions beyond project updates? | 17:27 |
| smcginnis | TheJulia: I think Jimmy was planning on finalizing an initial Forum schedule draft today. | 17:29 |
| TheJulia | smcginnis: ahh, good to know | 17:31 |
| smcginnis | TheJulia: And as I said that, just got a draft in my inbox. Should be final soon then. | 17:31 |
| TheJulia | smcginnis: \o/ | 17:33 |
| *** dtantsur is now known as dtantsur|afk | 17:38 | |
| *** annabelleB has quit IRC | 17:46 | |
| *** jamesmcarthur has joined #openstack-tc | 17:48 | |
| *** jamesmcarthur has quit IRC | 17:53 | |
| *** cdent has quit IRC | 17:54 | |
| *** diablo_rojo has joined #openstack-tc | 17:54 | |
| *** jamesmcarthur has joined #openstack-tc | 17:57 | |
| kmalloc | so, i just saw a bunch of "change py35 to py37 on stien+", i worry that if we drop testing for py35, we might sneak in py36/37-specific code. i am unaware of our overall python version targets. | 18:09 |
| kmalloc | i wanted to raise this with the TC to make sure we are properly communicating python versions and ensuring we aren't adding in incompatible code with supported versions of the code. | 18:10 |
| kmalloc | example: https://review.openstack.org/610687 | 18:10 |
| *** annabelleB has joined #openstack-tc | 18:12 | |
| kmalloc | so tentative -1 on those mechanical changes for keystone until the overall python version targets are clarified for me. | 18:12 |
| zaneb | kmalloc: so that change is wrong to me, because we explicitly said we were not going to drop 3.5 | 18:12 |
| kmalloc | zaneb: cool. | 18:12 |
| kmalloc | that confirms my view. | 18:12 |
| kmalloc | don't get me wrong, i'd love to get 36/37 only code | 18:13 |
| * zaneb looks for the reference | 18:13 | |
| kmalloc | the annotations are amazing | 18:13 |
| zaneb | right, but we don't know yet what will be in centos8 | 18:13 |
| zaneb | kmalloc: reference: https://governance.openstack.org/tc/goals/stein/python3-first.html#python-3-6-unit-test-jobs | 18:14 |
| zaneb | "We do not plan to update the minimum version of python 3 we support as part of this goal. Projects already running python 3.5 jobs should continue to do so." | 18:14 |
| kmalloc | ++ | 18:14 |
| kmalloc | hitting these with -2 | 18:14 |
| kmalloc | thanks | 18:14 |
| TheJulia | kmalloc: thanks! | 18:15 |
| kmalloc | TheJulia: i am hitting the keystone ones with -2 (and olso things we control) | 18:16 |
| *** annabelleB has quit IRC | 18:16 | |
| kmalloc | buit wanted to be sure everyone saw it. | 18:16 |
| * TheJulia nods | 18:16 | |
| fungi | zaneb: kmalloc: how are we not dropping py35? unless we're going to declare not-latest lts of ubuntu our new target for stein? | 18:17 |
| fungi | dropping py35 had nothing to do with python3 work, and everything to do with switching our default testing platform from xenial to bionic in stein | 18:17 |
| kmalloc | fungi: i know we can install py35 on a 18.04 base, and can help to do so. | 18:17 |
| kmalloc | so my concern is really the material changes in the python language between 35 and 35+ | 18:18 |
| kmalloc | there is a number of things that are incompat, and if we drop py27 near-ish term (iirc will happen) | 18:18 |
| kmalloc | it means we must also say py35 is not compat. | 18:18 |
| zaneb | fungi: many projects don't even support 3.6 yet, so if we drop 3.5 now there's no guarantee that we can even ship an OpenStack that runs on a single python | 18:18 |
| fungi | up to now we haven't declared support for specific python versions. we've declared support for specific platforms | 18:18 |
| fungi | zaneb: should we give up supporting latest ubuntu lts then? | 18:19 |
| * TheJulia feels really sad that we're having this discussion and some distributions still only package python 3.4 | 18:19 | |
| zaneb | it is a goal for Stein to get everyone onto 3.6 | 18:19 |
| kmalloc | i think we should declare python version support. i know i'm not on the TC, but that feels more correct than "distro" support. | 18:19 |
| fungi | TheJulia: which distributions package python 3.4? we declare that we will make sure openstack works on latest rhel (we substitute centos for licensing reasons) and latest ubuntu lts | 18:20 |
| kmalloc | just like we need to revisit mordred's patch for mysql version support | 18:20 |
| kmalloc | etc | 18:20 |
| kmalloc | we need to be better about delcaring minimums. | 18:20 |
| TheJulia | fungi: centos7 | 18:20 |
| fungi | TheJulia: centos doesn't package python3 at all | 18:20 |
| * kmalloc feels this is a bigger discussion | 18:20 | |
| kmalloc | fungi: i can remove the -2's when i am back from lunch | 18:20 |
| TheJulia | kmalloc: indeed | 18:21 |
| kmalloc | if we really need to move py35->py37 | 18:21 |
| fungi | kmalloc: if we're going to declare support for specific python versions, we should be building our own from source. distros have their own non-upstream patches in their python packages | 18:21 |
| TheJulia | fungi: I feel like I've been able to pull 3.4 out of their updates repo, but its fuzzy right now in my memory | 18:21 |
| * TheJulia blames cold meds | 18:22 | |
| fungi | we don't need to move from 3.5 to 3.7... we need to move from 3.5 (xenial) to 3.6 (bionic) but there's also a package of 3.7 in bionic some people want to test against for some additional safety | 18:22 |
| kmalloc | and we have 36 testing | 18:22 |
| kmalloc | so, i would like to set forth the timeline we support a python version, if it is "while the testing platform is bionic" we set the timeline for that | 18:23 |
| fungi | the overlap in testing with 3.5 and 3.6 was supposed to be temporary while we were dropping jobs on xenial and adding jobs on bionic | 18:23 |
| kmalloc | or such, but declaring the minimum python for openstack (similar to other packaged deps) really is important and i'm surprised we have gotten away with this as long as we have | 18:23 |
| mordred | kmalloc: well, so far the min has been "2.7" | 18:24 |
| kmalloc | mordred: right but with us moving to py3 consistently, i think we need to represent a minimum for py3 | 18:24 |
| lbragstad | kmalloc we should let coreyb know then? | 18:24 |
| mordred | totes | 18:24 |
| fungi | https://governance.openstack.org/tc/reference/project-testing-interface.html#linux-distributions | 18:24 |
| mordred | my understanding was that min for py3 was 3.5 | 18:24 |
| lbragstad | he has changes up to several repositories for py37 | 18:25 |
| kmalloc | mordred: that was my understanding as well, which is why i hit the patches owned by keystone with -2s | 18:25 |
| TheJulia | mordred: likewise | 18:25 |
| kmalloc | i | 18:25 |
| mordred | https://governance.openstack.org/tc/reference/pti/python.html <-- lists 3.5 specifically | 18:25 |
| kmalloc | 'll be back post lunch | 18:25 |
| kmalloc | and will followup. | 18:25 |
| kmalloc | :) | 18:25 |
| fungi | i think the expectation was we would update https://governance.openstack.org/tc/reference/pti/python.html whenever the default python on those platforms changed | 18:25 |
| kmalloc | sorry to raise the flag and run off. | 18:25 |
| * mordred throws bunnies at kmalloc | 18:25 | |
| kmalloc | mordred: i see you got the keys to the zoo back. | 18:25 |
| fungi | none of this is chiseled in stone for the release under development | 18:25 |
| * mordred is super excited about bumping that to 3.6 fwiw | 18:26 | |
| *** coreycb has joined #openstack-tc | 18:26 | |
| coreycb | lbragstad: hi o/ | 18:26 |
| lbragstad | discussion started here - http://eavesdrop.openstack.org/irclogs/%23openstack-tc/%23openstack-tc.2018-10-15.log.html#t2018-10-15T18:09:05 | 18:26 |
| mordred | and would TOTALLY support bumping it to 3.7 and dropping support for all other versions of python - but that's just me saying crazy words | 18:26 |
| zaneb | mordred: baby steps | 18:27 |
| coreycb | i think dhellmann is out on leave, and i've only had back-channel discussions with him on this | 18:27 |
| zaneb | it seems like we're running two parallel processes that are out of sync | 18:28 |
| coreycb | zaneb: yes i can see that, the py37 bits are new and not part of the python3-first initiative | 18:29 |
| zaneb | on one hand, the goals process that is trying to get everyone to 3.5 + 3.6 by the end of Stein; and on the other infra's regular switch to the latest Ubuntu LTS | 18:29 |
| lbragstad | ^ that's why i asked coreycb about the python3.7 stuff as it pertains to the community goal | 18:29 |
| fungi | zaneb: infra is not "regularly" switching any longer. not their responsibility to make sure openstack lives up to its claims as to what platforms it tests on | 18:29 |
| coreycb | lbragstad: it would be good to get a +1 from the tc. how would that be done? | 18:30 |
| fungi | if we, as openstack, fail to transition testing off xenial and onto bionic this cycle that's an openstack problem not an infra problem | 18:30 |
| lbragstad | coreycb well - this is a good first step | 18:30 |
| lbragstad | that or raising this during office-hours would be a good idea, too | 18:31 |
| fungi | a good second step would be to take this discussion to the ml | 18:31 |
| lbragstad | +1 ^ | 18:31 |
| lbragstad | arguable a good first step, too | 18:31 |
| coreycb | ok there's already a thread there but the subject may need updating to mention dropping of py3.5 | 18:31 |
| zaneb | fungi: ok, so let me amend that second bit to "OpenStack's stated goal to support the latest Ubuntu LTS" | 18:31 |
| fungi | because the visibility of this irc channel is significantly less than that of the -dev ml | 18:31 |
| clarkb | note that part of the motivation for disconnecting infra from the update process was the pain involved in doing the xenial switch. Many projects were unhappy about it | 18:32 |
| clarkb | I'd prefer that the project(s) own it themselves as a result | 18:32 |
| fungi | i'm perfectly fine if we decide we're going to to target ubuntu previous lts instead of current lts, but may make things increasingly harder for people who are trying to install/package for "modern" distributions | 18:33 |
| zaneb | so am I right in thinking that if we switch 3.6 jobs to Bionic and keep 3.5 jobs on Xenial, there is no conflict between these two processes? | 18:33 |
| fungi | the 3.6 jobs are already on bionic | 18:33 |
| zaneb | even better :) | 18:33 |
| fungi | what python version we use for jobs today is generally selected by proxy, choosing a particular distro and using the python it supplies | 18:34 |
| cmurphy | this discussion was already started on the ml fwiw http://lists.openstack.org/pipermail/openstack-dev/2018-October/135626.html | 18:34 |
| cmurphy | kmalloc: ^ | 18:34 |
| fungi | so, yes, if we can get everyone who is running python3.5-based jobs to also run then on 3.6 then we meet the requirements we've established, but at the expense of running twice as many jobs as we used to | 18:35 |
| clarkb | right and 3.7 is of interest because it changes enough stuff that things don't work and various individuals have noticed and asked that we test it more explicitly so we aren't caught unawares later | 18:35 |
| zaneb | not sure I agree it's chosen by proxy. all of the job names include the python version, as mordred pointed out https://governance.openstack.org/tc/reference/pti/python.html explicitly lists the python versions we support, and updating to a new python version has been a community-wide goal on two occasions | 18:35 |
| clarkb | zaneb: its been based on the distros we support | 18:36 |
| fungi | the reason the job names include py3x today is a reflection of the option that's passed to tox -e, but it still needs to be the python provided by the selected distro image | 18:37 |
| fungi | or we need to start building our own python | 18:38 |
| * mordred happily uses pyenv locally - but that's a much larger can of worms | 18:38 | |
| mordred | yah, it's at least bounded by the distros we support. like - I think we have the python3 min of 3.5 set because of xenial, but once things have migrated to bionic I could see bumping the py3 min to 3.6 | 18:38 |
| * zaneb has all of the pythons installed in Fedora | 18:39 | |
| mordred | alternately, we could choose to bump the min and by-proxy need to change the distro support policy - it's all policies under TC control after all :) | 18:39 |
| fungi | i personally build python from source at most recent tags for 3.4.x, 3.5.x, 3.6.x, 3.7.x and master (for future 3.8) and test changes for my own software against all of them (3.3 is eol and unable to build on debian sid due to needing older libraries than are available), but that would be a lot of jobs for openstack _and_ is no guarantee that testing against our own python builds is representative of | 18:39 |
| fungi | how the code will run on distro-patched-and-packaged python versions | 18:39 |
| mordred | fungi: I seem to remember backports of py3 things into py2 in distros a while back | 18:39 |
| clarkb | fungi: right and historically we've seen multiple cases where distro python differed from python python and not testing on it would've led to openstack being completely broken on $distro | 18:40 |
| mordred | yup | 18:40 |
| clarkb | mordred: that and the broken python3 on ubuntu multiple times in the past | 18:40 |
| mordred | yup | 18:40 |
| fungi | if we recommend users build their own python from source and run openstack with that, then testing with python built from source makes sense. if we recommend users install openstack on a distro and run it with the distro's packaged python interpreter, i think that's also what we should test | 18:40 |
| clarkb | its actually valuable to test what people will run with | 18:40 |
| coreycb | i would love to see more proactive testing on newer pythons. i'm sure python3.8 will be here soon | 18:40 |
| mordred | part of the issue is that we have folks with unittests that take 30-40 minutes - so a matrix of 4 versions of python3 becomes sort of ludicrously expensive | 18:41 |
| clarkb | I'm also no sure there is much value in testing intermediate pythons | 18:42 |
| zaneb | coreycb: yeah, the original sin here is that we were way too slow in getting to 3.6 | 18:42 |
| clarkb | particularly while we support python2 | 18:42 |
| coreycb | zaneb: yes but we can catch up ! :) | 18:42 |
| mordred | for those with quick unittests like keystoneauth and openstacksdk, one could do a single job that tests 2.7,3.5,3.6,3.7 in the same amount of time as a single unittest run for some projects | 18:42 |
| clarkb | test the boundaries and the stuff in the middle will largely be ok because python2 prevents you from using the fancy new features anyway | 18:42 |
| mordred | maybe we should put together a job that _can_ do a 4-in-one that people could use if they're interested in the coverage and it wouldn't otherwise kill them to have unittets run 4x as long | 18:43 |
| mordred | ? | 18:43 |
| mordred | clarkb: but yes - I also agree with that - testing only 3.7 seems fine if 2.7 is also being tested, because the surface area of "will be broken in 3.5 but work on both 2.7 and 3.7" is microscopic | 18:44 |
| mordred | but now we're getting in to a big mush area | 18:44 |
| clarkb | coreycb: has the python release cadence picked up speed? seems like precise was 3.3. trusty was 3.4. Xenial 3.5 and bionic 3.6 but now less than a year after bionic release we are talking 3.6? | 18:44 |
| clarkb | er talking 3.8 | 18:44 |
| mordred | clarkb: it does seem like it has increased doesn't it? | 18:44 |
| coreycb | clarkb: i'd have to check but i have heard something about 3.8 for 20.04 | 18:45 |
| zaneb | probably because python core team is not paying much attention to 2.7 any more, they can move faster | 18:45 |
| coreycb | https://docs.python.org/dev/whatsnew/3.8.html | 18:45 |
| coreycb | hey the good news is https://docs.python.org/dev/whatsnew/4.0.html is a 404 so far | 18:47 |
| mordred | \o/ | 18:49 |
| clarkb | if we want a minimum that doesn't follow particular release cadence of a distro then 3.5 seems reasonable as you get a bunch of the new good stuff that you can't use in python2 yet while still having access to things like the dict rewrite iirc | 18:50 |
| clarkb | then when python2 support is dropped you can make use of newer features without major rewrites or worrying about old python3 | 18:50 |
| coreycb | is it worth polling distros to see what versions they need, at least for stein? | 18:51 |
| clarkb | possibly? though on stein you can still python2 as a fallback | 18:51 |
| clarkb | train might be the important one for python3 version support and testing | 18:52 |
| coreycb | clarkb: right, but at least in our case we won't be using 3.5 for stein. 2.7, 3.6 and 3.7 will be our options. | 18:52 |
| coreycb | actually I think we are dropping py2 packages in stein. hopefully. | 18:53 |
| smcginnis | coreycb: No, the plan was to prepare in stein, then look at T to drop. We can't drop 2 in stein. | 18:58 |
| coreycb | smcginnis: sorry i wasn't specific enough, i'm talking about ubuntu specifically. | 18:58 |
| smcginnis | FWIW, I though the whole drop 3.5, pick up 3.6 adn 3.7 was all discussed on the ML thread a couple weeks ago by now we few objections. | 18:58 |
| smcginnis | coreycb: Oh, sorry. Was just trying to speed read to catch up on all the chatter over my lunch break. | 18:59 |
| coreycb | smcginnis: yes there's a thread, cmurphy linked it above | 18:59 |
| coreycb | smcginnis: no problem :) | 18:59 |
| zaneb | coreycb: replied on the mailing list | 19:00 |
| coreycb | zaneb: thanks | 19:02 |
| openstackgerrit | Corey Bryant proposed openstack/governance master: Add optional python3.7 unit test enablement to python3-first https://review.openstack.org/610708 | 19:24 |
| *** mriedem has quit IRC | 19:32 | |
| *** mriedem has joined #openstack-tc | 19:35 | |
| TheJulia | coreycb: But someone will see that 404 in the logs, and think it is a suggestion.... *blink* *blink* :) | 19:36 |
| coreycb | TheJulia: oops sorry :) | 19:36 |
| mordred | coreycb: ++ to dropping py27 packages in ubuntu | 19:58 |
| mordred | coreycb: if someone wants to 'apt-get install openstack-nova' - it seems completely legit for that to only install nova with python3.7 - and a great way to get people on to 3 who otherwise might be scared off | 19:59 |
| mordred | smcginnis: oh wow - we're actually considering dropping 2 in T? that would make me SO happy | 20:00 |
| smcginnis | mordred: I'm looking forward to when we can actually do that too. | 20:00 |
| mordred | smcginnis: especially if we can be on 3.6 as a min when we do | 20:00 |
| mordred | cause 3.6 has some really nice bits it would be wonderful to be able to use | 20:01 |
| *** jamesmcarthur has quit IRC | 20:13 | |
| smcginnis | TheJulia: https://www.openstack.org/summit/berlin-2018/summit-schedule/#track=262 | 20:26 |
| notmyname | if I have a question about the berlin schedule, who should I talk to? | 20:48 |
| TheJulia | notmyname: jamesmcarthur most likely... | 20:50 |
| notmyname | thanks | 20:50 |
| clarkb | notmyname: there is adev list thread too | 20:51 |
| notmyname | thanks. I found a summit@ email address | 20:57 |
| *** annabelleB has joined #openstack-tc | 20:58 | |
| *** annabelleB has quit IRC | 21:02 | |
| *** jamesmcarthur has joined #openstack-tc | 21:16 | |
| *** annabelleB has joined #openstack-tc | 21:27 | |
| kmalloc | coreycb, zaneb: removed my -2 from the keystone py3.7 patches. i am confident we'll get this covered in the TC resolutioin. | 21:38 |
| kmalloc | and followups from there | 21:38 |
| kmalloc | cheers and thanks! | 21:38 |
| kmalloc | mordred: i would love ot be on just py36 [or just py37] | 21:38 |
| kmalloc | mordred: i would *love* some of the annotation bits and that dicts are guaranteed ordered, etc, etc. | 21:39 |
| * kmalloc still thinks we should be calling out [even if it's driven by testing infrastructure] the binary minimums (and in the case of python 3x minmum even if we still say 27 is minimum python version) | 21:40 | |
| coreycb | kmalloc: thanks. I believe we've workflow -1'd all reviews until the tc review is resolved so that will block them for now. | 21:40 |
| clarkb | kmalloc: the annotations are in 3.5 iirc | 21:40 |
| kmalloc | coreycb: yup. | 21:40 |
| smcginnis | kmalloc: It does seem like we should publish those official versions per release somewhere. Maybe somewhere on releases.openstack.org. | 21:41 |
| kmalloc | clarkb: the annotations are, but in 36 there is expanded annotations that are super super super nice | 21:41 |
| kmalloc | smcginnis: ++++++++ | 21:41 |
| clarkb | and ordered dicts seem wrong even if cpython supports them. They could jsut as easily change it yet again | 21:41 |
| kmalloc | clarkb: in 37, the language spec is that dicts are ordered | 21:41 |
| kmalloc | in cpython 3.6 they happen to be | 21:41 |
| clarkb | kmalloc: ah ok I know that the implementation changed to make it the case but didn't know the lang spec says it now too | 21:42 |
| *** jamesmcarthur has quit IRC | 21:42 | |
| kmalloc | yeah :) | 21:42 |
| kmalloc | since the implementation in 36 was a good change(tm), it was included as part of the spec for 37 aiui | 21:42 |
| kmalloc | it doesn't cost a lot of extra mem to track the order of keys/values, but still get fast hash-table lookups. | 21:42 |
| clarkb | kmalloc: that wasn't the issue that made python3 change to unordred by default. THe problems were security related not performance. In particular determining hash collisions through lookup times and what they could tell you about random seeds iirc | 21:43 |
| kmalloc | ah | 21:47 |
| *** openstackgerrit has quit IRC | 21:56 | |
| *** openstackgerrit has joined #openstack-tc | 21:57 | |
| *** openstackgerrit has quit IRC | 21:58 | |
| *** openstackgerrit has joined #openstack-tc | 22:00 | |
| *** openstackgerrit has quit IRC | 22:12 | |
| *** openstackgerrit has joined #openstack-tc | 22:14 | |
| openstackgerrit | Corey Bryant proposed openstack/governance master: Add optional python3.7 unit test enablement to python3-first https://review.openstack.org/610708 | 22:18 |
| openstackgerrit | Corey Bryant proposed openstack/governance master: Add optional python3.7 unit test enablement to python3-first https://review.openstack.org/610708 | 22:18 |
| *** annabelleB has quit IRC | 22:32 | |
| *** openstackgerrit has quit IRC | 22:43 | |
| *** mriedem has quit IRC | 23:00 | |
| *** dklyle has quit IRC | 23:04 | |
| *** tosky has quit IRC | 23:25 | |
| *** dklyle has joined #openstack-tc | 23:30 | |
| *** jamesmcarthur has joined #openstack-tc | 23:31 | |
| *** jamesmcarthur has quit IRC | 23:35 | |
| * ttx still in China, meeting with various companies today (including around the office hour time) -- will discuss further developer engagement | 23:52 | |
| *** annabelleB has joined #openstack-tc | 23:55 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!