| *** david-lyle has quit IRC | 00:24 | |
| *** dims has quit IRC | 00:33 | |
| *** dims has joined #openstack-relmgr-office | 00:46 | |
| *** dims has quit IRC | 01:50 | |
| *** david-lyle has joined #openstack-relmgr-office | 02:08 | |
| *** dims has joined #openstack-relmgr-office | 03:46 | |
| *** dims has quit IRC | 03:51 | |
| *** redrobot has quit IRC | 03:55 | |
| ttx | dhellmann: I think we can still use "number of integrated releases" as the way to calculate the initial version number, considering pre-integration as 0.x basically | 06:37 |
|---|---|---|
| * ttx grabs coffee before entering office hours | 07:59 | |
| *** isviridov_away is now known as isviridov | 08:00 | |
| ttx | stevebaker, johnthetubaguy, SergeyLukjanov: I'm available to discuss new server versioning, in case you have concerns about that | 08:06 |
| ttx | (read http://lists.openstack.org/pipermail/openstack-dev/2015-June/067006.html first) | 08:08 |
| johnthetubaguy | ttx: 12.0.0 for nova seems good | 08:34 |
| johnthetubaguy | ttx: I know there was talk around the milestones, but I think thats more a case of making it clear what you get and what you don't | 08:35 |
| ttx | cool, just making sure you are aware of the change. We'll discuss it at cross-project meeting today, but two pings are better than one | 08:35 |
| johnthetubaguy | ttx: thats cool | 08:35 |
| ttx | as far as liberty-1 is concerned, we'll just tag 12.0.0b1 | 08:35 |
| ttx | i.e. s/2015.2.0/12.0.0/ | 08:35 |
| johnthetubaguy | ttx: I think thats the best choice right now, sounds good | 08:36 |
| johnthetubaguy | happy to see us make the move in time for the liberty-1 milestone | 08:36 |
| ttx | basaiclly we open the door for more independent releases, but expect most to still use the milestone pre-versioning system for liberty | 08:36 |
| johnthetubaguy | cool | 08:36 |
| ttx | denormalizing the release numbers will make the few that jump the shark a lot less funny-looking | 08:37 |
| ttx | better to do that in one go | 08:37 |
| johnthetubaguy | ttx: very true, its a nice idea | 08:37 |
| johnthetubaguy | ttx: I suspect for M we might want something that looks more like a full release, but waiting to see how ironic goes really | 08:37 |
| *** AzherKhan has joined #openstack-relmgr-office | 08:39 | |
| SergeyLukjanov | ttx, I think 3.0.0 is ok for Sahara (due to the there were 0.1, 0.2, 0.3, Icehouse, Juno and Kilo releases) | 09:18 |
| SergeyLukjanov | ttx, good morning | 09:18 |
| ttx | SergeyLukjanov: cool | 09:18 |
| *** dims has joined #openstack-relmgr-office | 09:18 | |
| SergeyLukjanov | ttx, oh, I'm just understand that I was counting from zero | 09:23 |
| SergeyLukjanov | ttx, so, we have 2014.1, 2014.2 and 2015.1 tags already | 09:23 |
| *** dims has quit IRC | 09:23 | |
| SergeyLukjanov | ttx, in etherpad all 2015.1 versions are missed | 09:24 |
| SergeyLukjanov | ttx, as it as designed? | 09:24 |
| ttx | SergeyLukjanov: hmm, probably not | 09:25 |
| SergeyLukjanov | ttx, probably for sahara it'll be better to tag 4.0.0 (0.X for the tags 0.1, 0.2, 0.3; 1.X for 2014.1, 2.X for 2014.2 and 3.X for 2015.1) | 09:25 |
| SergeyLukjanov | it seems more logical for me | 09:25 |
| ttx | yes, will cross-check with dhellmann | 09:25 |
| SergeyLukjanov | ttx, yup, I'll post it to ML and I'll be on cross-project meeting | 09:26 |
| ttx | SergeyLukjanov: actually 2014.2 was the first "integrated" version | 09:35 |
| SergeyLukjanov | ttx, yeah, but should we skip non-integrated versions? | 09:35 |
| ttx | SergeyLukjanov: so we could consider it's 1.0 | 09:36 |
| SergeyLukjanov | ttx, it's an option, yeah | 09:36 |
| ttx | SergeyLukjanov: I'll rediscuss the ruls with dhellmann | 09:36 |
| ttx | rule* | 09:36 |
| ttx | I think that's what makes the most sense (consider pre-integrated as 0.x and count from 1.0 for integrated) | 09:37 |
| ttx | I'm editing the etherpad to match | 09:37 |
| SergeyLukjanov | ttx, ack | 09:37 |
| SergeyLukjanov | ttx, so, for the Liberty release we'll have a list of corresponding project versions? | 09:42 |
| ttx | yes | 09:42 |
| *** AzherKhan has quit IRC | 10:46 | |
| *** dims has joined #openstack-relmgr-office | 11:21 | |
| *** dims has quit IRC | 11:25 | |
| *** dims has joined #openstack-relmgr-office | 11:29 | |
| *** redrobot has joined #openstack-relmgr-office | 12:42 | |
| *** redrobot is now known as Guest74993 | 12:42 | |
| *** Guest74993 is now known as el_robot_rojo | 12:43 | |
| *** el_robot_rojo is now known as redrobot | 12:44 | |
| dhellmann | ttx: hi | 13:03 |
| * dhellmann tries to figure out why his script missed the 2015 tags | 13:03 | |
| dhellmann | ah, those are 2015.1.0 and didn't match my regex | 13:05 |
| ttx | anyway, all in all that makes most numbers unchanged :) | 13:08 |
| ttx | (with my suggestion of considering pre-integration as 0.x) | 13:08 |
| ttx | Also I added to the etherpad what would be missing tags in the git history | 13:08 |
| ttx | like the 2010.1 Nova Austin | 13:09 |
| dhellmann | yeah, I guess we can count pre-integration releases as 0.x, I hadn't thought of it that way | 13:10 |
| dhellmann | so projects that have never been integrated will start at 1.0? | 13:11 |
| openstackgerrit | Doug Hellmann proposed openstack-infra/release-tools: Add show_legacy_releases.sh https://review.openstack.org/191228 | 13:11 |
| ttx | yeah they would | 13:11 |
| dhellmann | ok | 13:11 |
| ttx | that is what I posted in my ML answer | 13:12 |
| ttx | That will make a designate 1.0.0 which may look funny but I can live with it | 13:12 |
| ttx | now is as good a time as any | 13:12 |
| ttx | and frankly if they really want to start at 2.0.0 minimum for whatever marketing reason, I don't really care | 13:13 |
| dhellmann | ttx: the other projects you list there aren't tagged as release:managed in the governance repo, so that's why I didn't include them | 13:13 |
| dhellmann | barbican, designate, etc. | 13:13 |
| ttx | yes, I know. In the original submission of release:managed I left them out so that we still had the option to not support them if we really didn't want that | 13:14 |
| dhellmann | ok | 13:14 |
| ttx | at least until we had a team discussion about them | 13:14 |
| ttx | but I managed their kilo releases, so I guess we could/should manage their liberty ones | 13:14 |
| ttx | unless we feel it's too much work | 13:15 |
| ttx | If we agree, I can push a release:managed tag addition there | 13:15 |
| dhellmann | ok, I don't have a problem with that if the teams don't | 13:15 |
| ttx | I'll ask for their feedback on the governance proposal | 13:15 |
| dhellmann | I will also need to re-update the ACLs patch for the libraries from those projects | 13:15 |
| ttx | or we can make an additional dependent patch | 13:16 |
| dhellmann | I'll probably do that | 13:17 |
| dhellmann | ttx: did you have any thoughts about the neutron plugins? | 13:19 |
| ttx | the -aaS things ? | 13:24 |
| ttx | I commented on the ML as well -- they were released in lockstep before, so reusing the neutron version sounds fair there | 13:24 |
| dhellmann | ok, I guess that works. I thought they were newer than neutron, though, so I didn't want to assume the version would be the same | 13:25 |
| ttx | dhellmann: At this point mestery always considered those parts of the same "neutron" deliverable | 13:27 |
| dhellmann | ok, cool | 13:27 |
| ttx | i.e. you're not supposed to use different versions and expect that to work | 13:28 |
| ttx | I suspect that may evolve in the future | 13:28 |
| ttx | dhellmann: I managed to reach Sergey and John G. during the office hours this morning. Both agreed on their versions. | 13:47 |
| dhellmann | ttx: are we sprinting on the project guide today? | 13:47 |
| ttx | that would be Thursday/Friday | 13:47 |
| * dhellmann checks why his alarm went off early | 13:47 | |
| * dhellmann might need to just go back to bed today | 13:47 | |
| * ttx takes a break | 13:48 | |
| *** isviridov is now known as isviridov_away | 14:33 | |
| SlickNik | ttx: Hi — I had a topic to discuss to the cross-project meeting today. Is it too late to add it to the agenda? | 16:31 |
| SlickNik | It is regarding a follow up on http://lists.openstack.org/pipermail/openstack-dev/2015-June/065731.html | 16:33 |
| dhellmann | SlickNik: that looks like something to bring up in the "vertical projects" open discussion period | 16:38 |
| SlickNik | dhellmann: Sounds good. I'll plan on doing that. | 16:39 |
| dhellmann | SlickNik: great, see you then | 16:40 |
| *** stevemar has joined #openstack-relmgr-office | 17:17 | |
| * stevemar is standing in for / helping morganfainberg | 17:17 | |
| stevemar | dhellmann, proposed server versions of oslo? | 17:18 |
| dhellmann | stevemar: we're re-versioning all of the server projects, like keystone, to drop date-based versioning | 17:19 |
| dhellmann | http://lists.openstack.org/pipermail/openstack-dev/2015-June/067082.html | 17:19 |
| dhellmann | and https://etherpad.openstack.org/p/liberty-initial-semver-values | 17:19 |
| stevemar | oh the whole 2015.1 ... and so forth | 17:19 |
| dhellmann | yep | 17:19 |
| mestery | dhellmann: So, the tl;dr on the neutron *aas projects is that they will version DIFFERENTLY than the server? I tend to agree with ihar's comments that they should version the same since we don't have separate API endpoints for htem yet. | 17:19 |
| stevemar | oh gosh yes | 17:19 |
| dhellmann | stevemar: I'm going to be submitting patches using this version numbers tomorrow | 17:19 |
| dhellmann | mestery: no, ttx convinced me it's ok to just use the same values so they'll all start with 7.0.0 | 17:20 |
| dhellmann | mestery: sorry, that thread got confusing because I was participating in a state of undercaffination | 17:20 |
| mestery | lol | 17:20 |
| mestery | dhellmann: OK, you've got my ack on keeping them all the same, thank you for driving this! | 17:20 |
| dhellmann | mestery: thanks! | 17:20 |
| stevemar | dhellmann, i couldn't be happier with the versioning | 17:21 |
| stevemar | i really didn't like the date based approach :\ | 17:21 |
| dhellmann | cool, so we'll go with keystone as 8.0.0 | 17:21 |
| stevemar | makes sense to me. liberty will be our 8th release | 17:21 |
| dhellmann | right, that's the idea | 17:21 |
| morganfainberg | dhellmann: I want version 13.0 (jk) | 17:21 |
| dhellmann | morganfainberg: exceptions can be made, but you have to earn them | 17:22 |
| * dhellmann sounds like he's selling indulgences | 17:22 | |
| morganfainberg | dhellmann: /put ptl card on table and stamps feet (does that count? :P | 17:22 |
| mestery | lol | 17:22 |
| * dhellmann may have been watching too much Wolf Hall | 17:22 | |
| morganfainberg | 8.0 sounds good. | 17:22 |
| dhellmann | morganfainberg: excellent, thanks | 17:22 |
| stevemar | dhellmann, so if something goes awry, then it's 8.0.1 ? | 17:23 |
| dhellmann | or 8.1 or whatever, depending on what has merged | 17:23 |
| stevemar | cool cool | 17:23 |
| stevemar | keeps things more pythony | 17:23 |
| dhellmann | actually, we'll be doing pre-release versioning still, so L1 will be 8.0.0b1 I think | 17:23 |
| stevemar | ohhh | 17:23 |
| morganfainberg | Is that going to fubar pip up for install because 2015.x.x > 8.0.0? | 17:23 |
| stevemar | fancy | 17:23 |
| dhellmann | but *after* liberty, the stable releases will be semver | 17:23 |
| morganfainberg | Or whatever prev versions were. | 17:23 |
| morganfainberg | Oh wait nvm. | 17:24 |
| morganfainberg | deep. | 17:24 |
| dhellmann | morganfainberg: it will for upgrades, although apparently pip supports epochs, too | 17:24 |
| morganfainberg | Derp* | 17:24 |
| morganfainberg | We don't put keystone on pypi | 17:24 |
| dhellmann | well, that's part of the move, actually | 17:24 |
| morganfainberg | Because... Go look what keystone is there :P | 17:24 |
| stevemar | https://pypi.python.org/pypi/Keystone :P | 17:24 |
| morganfainberg | We don't own it. It is a "framework". | 17:24 |
| stevemar | that's totally us, i dunno what you're talking about | 17:25 |
| dhellmann | keystone is all things to all people | 17:25 |
| morganfainberg | I hear we are a key value store over rest. | 17:25 |
| morganfainberg | But implemented without the basic functionality of say memcache. | 17:25 |
| * morganfainberg stops being snarky and goes back to AFK. | 17:26 | |
| morganfainberg | dhellmann: actually. Hmm will get back to you shortly. | 17:26 |
| morganfainberg | Have a thought... | 17:26 |
| * dhellmann stands by for thought download | 17:29 | |
| morganfainberg | Nope. Lost it. :( | 17:51 |
| morganfainberg | Oh well. Thought is gone. | 17:51 |
| dhellmann | ttx: do any of the scripts in release-tools help with updating the version number in setup.cfg, or should I just do those by hand? | 18:24 |
| ttx | dhellmann: no, none | 18:52 |
| dhellmann | ttx: ok, it's just a few so I'll do them by hand | 19:10 |
| *** Daviey has quit IRC | 21:30 | |
| dhellmann | ttx: it looks like we need to tag all of the projects changing away from date-based releases with a version less than the new version they are working up to in order to make pbr happy. So I am planning to use X.0.0a0 as a tag for all of those projects. Then after we add the setup.cfg change, we will get versions like X.0.0a1.devN and then we can tag X.0.0a1 at L1 or we can just leave the tagging for later. | 21:55 |
| dhellmann | ttx: let me know if you like the a0 or if I should go ahead and use a1 for now or if you don't care one way or the other. :-) | 21:56 |
| lifeless | dhellmann: you're so much more polite than I... I'd have JFDI'd ;) | 21:56 |
| lifeless | dhellmann: am I accepted into the team at this point? I'm not in https://review.openstack.org/#/admin/groups/11,members | 21:56 |
| dhellmann | lifeless: I'm new on the team. :-) | 21:56 |
| lifeless | ttx: ^ | 21:57 |
| dhellmann | lifeless: oh, I was planning to add you to the new library-release team, but hadn't thought about that one. I'll defer to ttx as PTL | 21:57 |
| lifeless | oh | 21:57 |
| lifeless | didn't realise there were two | 21:57 |
| lifeless | whatever works for ttx | 21:57 |
| dhellmann | lifeless: yeah, the new team won't exist until the acl patch lands and gerrit creates it as a side-effect | 21:58 |
| dhellmann | or maybe fungi creates those after the patches land, but either way, after the patch | 21:58 |
| ttx | let's do library-release for now | 22:00 |
| dhellmann | ttx: ack | 22:01 |
| dhellmann | ttx: you can think about the version number thing tomorrow morning, I won't be doing that until my morning so there's time | 22:01 |
| ttx | dhellmann: on the tag thing | 22:01 |
| dhellmann | oh, unless you have an answer ready :-) | 22:01 |
| ttx | was wondering what would happen if we just tagged X.0.0b1 and then change setup.cfg | 22:02 |
| ttx | instead of X.0.0a1, setup.cfg; X.0.0b1 | 22:02 |
| dhellmann | oh, I wasn't using beta versions in the setup.cfg | 22:02 |
| dhellmann | so you mean tag b1 on next thursday? | 22:02 |
| ttx | right, you shouldn't use beta version in setup.cfg | 22:03 |
| dhellmann | and then after the tag, change the setup.cfg to have X.0.0? | 22:03 |
| ttx | yes | 22:03 |
| dhellmann | I think that would probably work. lifeless ? | 22:03 |
| ttx | I guess the problem is that you generate crappy 2015.2.0-foo-bar versions between X.0.0b1 and the setup.cfg change | 22:03 |
| ttx | since that wouldn't be atomic | 22:04 |
| dhellmann | let me test that | 22:04 |
| ttx | so it's a bit ugly, but wouldn't be the first time we generate a random crappy version | 22:04 |
| dhellmann | if I tag 1.0.0b1 and then commit the setup.cfg change I get 1.0.0.0b2.dev1 | 22:05 |
| ttx | also we could ask the PTLs to approve the setup.cfg bump first thing after we push the tag | 22:05 |
| ttx | dhellmann: right, so that would work as well | 22:05 |
| ttx | and avoid the extraneous a1 tag | 22:05 |
| dhellmann | that makes the timing a bit trickier, which might be a reason to live with the alpha tag, but either way should get us a good beta tag on thursday | 22:06 |
| ttx | If it's a technical tag we could (I think) remove it from history | 22:07 |
| dhellmann | that might break folks doing ci/cd | 22:07 |
| ttx | hmm, yeah | 22:07 |
| ttx | so option 1 is to tag 1.0.0a1 before setup.cfg and 1.0.0b1 tag, but results in extraneous tag | 22:08 |
| ttx | option 2 is to tag 1.0.0b1 and then bump setup.cfg | 22:08 |
| lifeless | sorry, stepped away for a sec | 22:08 |
| lifeless | reading | 22:08 |
| ttx | both cases we'll generate "wrong" versions for any commit between the tag and the setup.cfg bump | 22:08 |
| ttx | i.e. 2015.2.0-somethign-something after the 1.0.0a1 or 1.0.0b1 tag | 22:09 |
| lifeless | you can't have the same tag versionin setup.cfg as you've tagged. | 22:09 |
| ttx | lifeless: sure | 22:09 |
| lifeless | or are you saying 'wait until the beta is cut before updating setup.cfg ? | 22:10 |
| dhellmann | ttx: no, once we tag the 2015 versions go away | 22:10 |
| ttx | we would tag 1.0.0b1, then push 1.0.0 to setup.cfg | 22:10 |
| lifeless | so | 22:10 |
| lifeless | the thing thats weird there | 22:10 |
| ttx | dhellmann: not for intermediary commits | 22:10 |
| dhellmann | oh, hrm, maybe they only go away because I'm sitting on a tag -- yeah | 22:10 |
| lifeless | is that the first commit after the tag would be back in the 2015 space | 22:10 |
| ttx | lifeless: yes, that's what I expected | 22:10 |
| ttx | except if the next commit is the setup.cfg bump, right ? | 22:11 |
| lifeless | I think thats confusing, and it would make getting those first commits to be the setup.cfg change really critical | 22:11 |
| lifeless | I don't like that | 22:11 |
| ttx | lifeless: but you get the exact same effect if you follow option 1 | 22:11 |
| lifeless | its also confusing because the tagged version will have a wildly different version = line | 22:11 |
| lifeless | ttx: for a tag we're not announcing and not telling folk to use | 22:11 |
| lifeless | ttx: or test | 22:11 |
| lifeless | ttx: thats a big social difference IMO | 22:11 |
| ttx | you'll have a 2015.2 version for any commit between 1.0.0a1 tag and setup.cfg bump to 1.0.0 | 22:11 |
| ttx | lifeless: that's fair | 22:12 |
| lifeless | there is a way around this | 22:12 |
| lifeless | remove the version = from setup.cfg. Tag the new lower bound. Add version = back into setup.cfg | 22:12 |
| lifeless | but its more steps for marginal gain. | 22:12 |
| lifeless | and may involve two version sequence rewinds if the current version= in setup.cfg is higher than pbr would generate itself. | 22:13 |
| ttx | ok, so I'll jump to bed. I'm fine with a1 tags as a workaround. Would prefer if we did not add extraneous tags but I'm fine with it | 22:13 |
| dhellmann | removing the version from setup.cfg for barbican causes the version to go backwards from 2015.2.0.dev101 to 2015.1.1.dev101 | 22:14 |
| ttx | dhellmann: I think we shouldn't remove the setup.cfg version. Just use a1 tags | 22:14 |
| dhellmann | you know, another option is to do the tag and then just remove the entry from setup.cfg entirely and rely on tags from now on | 22:15 |
| dhellmann | I've so far not wanted to do that, because it's a much bigger procedural change | 22:15 |
| ttx | dhellmann: sounds risky. I prefer the a1 trick | 22:15 |
| dhellmann | ttx: agreed. a0 or a1? | 22:15 |
| ttx | whatever :) | 22:16 |
| dhellmann | the a0 tag will produce a1 versions for later commits | 22:16 |
| dhellmann | I guess it doesn't really matter, so maybe a1 is clearer as a tag | 22:16 |
| ttx | a0 sounds good for a null tag | 22:16 |
| ttx | your call | 22:16 |
| ttx | a0/a1 | 22:16 |
| * ttx disappears | 22:16 | |
| dhellmann | k | 22:16 |
| dhellmann | gnite | 22:16 |
| * dhellmann heads out for dinner | 22:17 | |
| ttx | good end of day! | 22:17 |
| *** openstackgerrit has quit IRC | 22:38 | |
| *** openstackgerrit has joined #openstack-relmgr-office | 22:39 | |
| SlickNik | ttx / dhellmann: Is the TC meeting next week on Monday (22nd) instead of Tue (23rd)? The reason I'm asking is because the wiki seemed to indicate that (https://wiki.openstack.org/wiki/Meetings/TechnicalCommittee) and I wanted to check it wasn't a typo. | 22:54 |
| SlickNik | Thanks! | 22:54 |
| *** sdague has quit IRC | 23:12 | |
| *** sdague has joined #openstack-relmgr-office | 23:24 | |
| *** stevemar has quit IRC | 23:44 | |
| *** stevemar has joined #openstack-relmgr-office | 23:44 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!