| *** brinzhang has joined #openstack-requirements | 00:21 | |
| *** vishalmanchanda has joined #openstack-requirements | 00:21 | |
| *** brinzhang_ has joined #openstack-requirements | 01:35 | |
| *** brinzhang_ has quit IRC | 01:37 | |
| *** brinzhang_ has joined #openstack-requirements | 01:37 | |
| *** brinzhang has quit IRC | 01:39 | |
| *** brinzhang has joined #openstack-requirements | 01:59 | |
| *** brinzhang has quit IRC | 02:00 | |
| *** brinzhang has joined #openstack-requirements | 02:02 | |
| *** brinzhang_ has quit IRC | 02:02 | |
| *** brinzhang_ has joined #openstack-requirements | 02:22 | |
| *** brinzhang has quit IRC | 02:25 | |
| *** brinzhang has joined #openstack-requirements | 02:44 | |
| *** brinzhang_ has quit IRC | 02:47 | |
| *** vishalmanchanda has quit IRC | 03:09 | |
| adriant | amotoki: this commit and the others like it, break installation for horizon from source: https://github.com/openstack/requirements/commit/6aa3b0ced466089cab5cf9eb529af2c1a4c67976#diff-0bdd949ed8a7fdd4f95240bd951779c8 | 03:13 |
|---|---|---|
| adriant | because I am trying to install our fork of horizon, with constraint to rocky, and because I am install horizon using pip from source, it breaks since that commit | 03:14 |
| adriant | and I've had to pin our upper-constraints to the commit RIGHT before that one | 03:14 |
| adriant | I can understand the requirements for plugins in horizon wanting an upper-constraint, but this breaks our build, and potentially others would just like to build from source and use the upstream 'stable' constraint branches | 03:15 |
| adriant | ERROR: Could not satisfy constraints for 'horizon': installation from path or url cannot be constrained to a version | 03:16 |
| adriant | with the install command being: | 03:17 |
| adriant | pip install -c https://raw.githubusercontent.com/openstack/requirements/stable/rocky/upper-constraints.txt /opt/horizon/ | 03:17 |
| adriant | and our source code for horizon sitting in /opt/horizon | 03:17 |
| adriant | that used to work before your commit | 03:19 |
| adriant | kind of related: https://github.com/pypa/pip/issues/6310 | 03:20 |
| prometheanfire | so, hi | 03:22 |
| prometheanfire | I don't like them being in tree as a constraint either, but their plugins using them as libs makes it impossible to avoid | 03:23 |
| prometheanfire | they are only constrained on stable branches | 03:23 |
| adriant | but that kind of breaks anyone doing builds from source like this :/ | 03:23 |
| adriant | and I'd assume that not an uncommon pattern | 03:23 |
| prometheanfire | packagers / deployers are welcome do use our upper-consrtaints as is, or modify them to their need | 03:23 |
| adriant | but then we're maintaining a fork of upperconstaints too | 03:24 |
| adriant | so many forks :( | 03:24 |
| prometheanfire | I would stress (as a packager myself) that constraints are a very strong suggestion, if you don't lik something, you can change it | 03:24 |
| prometheanfire | ya :( | 03:24 |
| prometheanfire | a pipe'd sed could work too | 03:24 |
| prometheanfire | less forky | 03:24 |
| adriant | maybe | 03:24 |
| prometheanfire | unfortunaltey it's just a bad situation | 03:25 |
| adriant | I would argue that we should solve this differently than breaking this for people | 03:25 |
| adriant | this is a BAD solution | 03:26 |
| adriant | not that I have a useful alternative :/ | 03:26 |
| adriant | the plugins only use horizon as a lib for testing | 03:32 |
| adriant | you can't exactly have plugins without horizon itself | 03:32 |
| adriant | I have the exact same problem with Adjutant, although i can't say I solve it, other than have the plugins rely on adjutant master repo for testing, and potentially will start pinning test-requirments in my plugins to specific stable branches of Adjutant in their own stable branches. | 03:33 |
| adriant | maybe that is a saner pattern to keep horizon out of upperconstraints? | 03:34 |
| adriant | and I guess we may have similar issues with: | 03:35 |
| adriant | https://github.com/openstack/requirements/commit/1d44f821536a970cd804abe31cba3bcd5188bcb1 | 03:35 |
| adriant | honestly this sounds like a nightmare. Adding services to upper-constraints shouldn't be allowed | 03:36 |
| adriant | or we need some official pattern to do it in a sane way | 03:36 |
| adriant | "Generally adding OpenSatck services isn't allowed but..." | 03:37 |
| * adriant sighs | 03:37 | |
| prometheanfire | adriant: it's been discussed, (sorry, was afk) | 03:37 |
| prometheanfire | the alternate (right) solution is to get the projects (neutron/horizon) to enforce the non-usage of themselves as libraries, forking library code to $PROJECT-lib and having both the libraries and plugins use that | 03:38 |
| prometheanfire | ya, it's a pain | 03:39 |
| adriant | like, if pip allowed more than one constraint file, we could split this out | 03:39 |
| adriant | or even have one with services, one without... | 03:39 |
| adriant | does mean duplication but doesn't break anyone | 03:40 |
| prometheanfire | possibly, I thought that it did, or that the last one specified took priority | 03:40 |
| adriant | does it allow more than one? | 03:40 |
| prometheanfire | I thought it did, but may be thinking of requirements files | 03:40 |
| adriant | because then you can have "upper-constraints.txt" and "service-upper-constraints.txt" or something, and places that need the latter, use both | 03:41 |
| adriant | with the latter having just services | 03:41 |
| adriant | that happen to be used as libs... | 03:41 |
| adriant | that's not exactly a good idea either | 03:42 |
| prometheanfire | right | 03:43 |
| prometheanfire | it's just shitty all around | 03:43 |
| prometheanfire | the project won't do anything to split out -lib versions | 03:43 |
| adriant | I'd hard enforce that. do a -lib library, or don't get added to upper-constraints | 03:44 |
| adriant | the alternative is kind of worse, and we're there now... | 03:44 |
| adriant | and I say this knowing I'll get hit by that issue myself :P | 03:45 |
| adriant | since Adjutant plugins rely on Adjutant itself for testing | 03:45 |
| prometheanfire | hmm I should have them do whatever causes https://github.com/pypa/pip/issues/5079 | 03:45 |
| prometheanfire | I might make that a cycle goal, might go like this | 03:46 |
| prometheanfire | hmm, just removing them again early in the cycle could be interesting | 03:47 |
| prometheanfire | not sure it's within this projects scope, (maintaning co-installable openstack), part of that definition is not caring about the outside (sorry adriant) | 03:48 |
| prometheanfire | the problem I've had in getting projects to do $FOO-lib is that the projects themselves don't necessarilly care if I (a third party project) ask, also, if I have them or don't in constraints doesn't mater to them either, it's all on the libs | 03:50 |
| adriant | I would assume being able to use upper-constraints to install stable version of openstack services with sane versions of making libraries IS well within the project scope | 03:50 |
| *** timburke has quit IRC | 03:50 | |
| prometheanfire | I guess this would force the libs to keep the -lib version updated | 03:50 |
| prometheanfire | define sane versions :D | 03:50 |
| adriant | versions semi-officially decided by the community and merged into this repo | 03:51 |
| prometheanfire | constraints are first and foremost for gating | 03:51 |
| adriant | I'm fairly sure we are not the only company using upper-constraints to install stable releases of the services | 03:51 |
| prometheanfire | I'm sure you are not | 03:52 |
| adriant | I'd assume thats a common and sane practice to ensure library compatibility with the version of the software we are trying to run | 03:52 |
| prometheanfire | yep | 03:52 |
| adriant | so maybe it should be considered as part of the scope of the project officially? | 03:52 |
| prometheanfire | I'm not seeing why you are having a problem really, lets get back to the basic question | 03:52 |
| adriant | "ERROR: Could not satisfy constraints for 'horizon': installation from path or url cannot be constrained to a version" | 03:53 |
| prometheanfire | is your problem that you cannot install a fork of horizon using upstream constraints? | 03:53 |
| adriant | yes | 03:53 |
| adriant | or even upstream horizon from source | 03:53 |
| adriant | fork doesn't matter | 03:53 |
| *** timburke has joined #openstack-requirements | 03:53 | |
| prometheanfire | it should install from pip | 03:53 |
| adriant | in a docker container, clone a horizon repo, install using pip and upper-constraint | 03:53 |
| prometheanfire | should be trying horizon===14.something right? | 03:54 |
| adriant | the problem is horizon's weird enabled files and such make it better handled as source | 03:54 |
| adriant | at least in my experience | 03:54 |
| prometheanfire | ah, zuul got aroud it by removing the project being installed from constraints, so horizon would install unconstrainted via zuul iirc | 03:54 |
| adriant | I'd rather and more commonly, install from source than pypi for services | 03:55 |
| prometheanfire | that's what it mentions with zuulv3, even openstack gate modifies the constraints file :P | 03:55 |
| adriant | not sure if other do the same or that's just use | 03:55 |
| * adriant sighs | 03:55 | |
| adriant | ok... I guess I could add that as a step in our pipeline | 03:56 |
| adriant | but it would be nice if we had a blacklist of such services in a file somewhere in this repo I could source to strip from the other | 03:56 |
| prometheanfire | ya, perfect world would be nice and I'll revisit the -lib thing with others soon (reqs meeting is wednesday) | 03:56 |
| adriant | less stuff we need to keep in sync | 03:56 |
| prometheanfire | pipeline would mean less to keep in sync (sed, etc) | 03:56 |
| adriant | well yeah, but I'd still need to maintain my own list of blacklist values | 03:57 |
| adriant | which may not change often mind you | 03:57 |
| adriant | but still :P | 03:57 |
| prometheanfire | the main blacklist is what we maintian for testing/top_level_projects | 03:57 |
| prometheanfire | https://github.com/openstack/requirements/blame/master/blacklist.txt | 03:57 |
| prometheanfire | ansible-lint six months ago was most recent | 03:58 |
| prometheanfire | I was able to head off sahara (see the commit message for that one) | 03:59 |
| prometheanfire | horizon/neutron were before my time :( | 03:59 |
| adriant | prometheanfire, well thanks for listening, and sorry for ranting. If you can bring it up and something sensible happens I'd be happy to know. Just annoyed because "this used to work", and now it doesn't without modification. | 04:03 |
| prometheanfire | I've had a rant or two | 04:03 |
| *** brinzhang_ has joined #openstack-requirements | 04:13 | |
| *** brinzhang_ has quit IRC | 04:15 | |
| *** brinzhang_ has joined #openstack-requirements | 04:15 | |
| *** brinzhang has quit IRC | 04:17 | |
| *** udesale has joined #openstack-requirements | 04:23 | |
| *** vishalmanchanda has joined #openstack-requirements | 04:31 | |
| *** brinzhang has joined #openstack-requirements | 05:04 | |
| *** brinzhang_ has quit IRC | 05:07 | |
| openstackgerrit | Merged openstack/requirements master: update constraint for python-openstackclient to new release 5.0.0 https://review.opendev.org/709591 | 05:31 |
| *** evrardjp has quit IRC | 05:34 | |
| *** evrardjp has joined #openstack-requirements | 05:35 | |
| *** brinzhang_ has joined #openstack-requirements | 05:37 | |
| *** brinzhang has quit IRC | 05:40 | |
| *** brinzhang_ has quit IRC | 06:02 | |
| *** strigazi has quit IRC | 06:16 | |
| *** strigazi has joined #openstack-requirements | 06:18 | |
| *** tosky has joined #openstack-requirements | 08:05 | |
| *** ccamacho has joined #openstack-requirements | 08:05 | |
| *** e0ne has joined #openstack-requirements | 08:24 | |
| *** ralonsoh has joined #openstack-requirements | 08:40 | |
| *** witek_ has joined #openstack-requirements | 08:46 | |
| *** iurygregory has joined #openstack-requirements | 08:50 | |
| *** hberaud has joined #openstack-requirements | 09:00 | |
| *** hberaud has quit IRC | 09:06 | |
| *** udesale has quit IRC | 09:07 | |
| *** hberaud has joined #openstack-requirements | 09:15 | |
| *** e0ne has quit IRC | 09:16 | |
| *** hberaud has quit IRC | 10:09 | |
| *** hberaud has joined #openstack-requirements | 10:10 | |
| *** hberaud has quit IRC | 10:22 | |
| *** hberaud has joined #openstack-requirements | 10:23 | |
| *** e0ne has joined #openstack-requirements | 10:32 | |
| *** brinzhang has joined #openstack-requirements | 10:36 | |
| *** e0ne has quit IRC | 11:16 | |
| *** udesale has joined #openstack-requirements | 11:17 | |
| *** iurygregory has quit IRC | 11:26 | |
| *** brinzhang_ has joined #openstack-requirements | 11:40 | |
| *** brinzhang has quit IRC | 11:43 | |
| *** brinzhang_ has quit IRC | 11:46 | |
| *** witek_ has quit IRC | 14:05 | |
| smcginnis | I do not like having services in upper constraints either. But I also do not think it's upstreams place to exclude those just so someone can install a fork of a service. That's entirely a downstream concern that has to be dealt with by maintaining its own upper constraints. | 14:05 |
| smcginnis | Don't install a fork. | 14:06 |
| smcginnis | (easy for me to say, I know) | 14:06 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: update constraint for oslo.log to new release 4.0.1 https://review.opendev.org/709731 | 14:23 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: update constraint for oslo.config to new release 8.0.1 https://review.opendev.org/709736 | 14:28 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: update constraint for oslo.rootwrap to new release 6.0.1 https://review.opendev.org/709737 | 14:29 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: update constraint for oslo.utils to new release 4.0.1 https://review.opendev.org/709738 | 14:30 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: update constraint for oslo.concurrency to new release 4.0.1 https://review.opendev.org/709739 | 14:32 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: update constraint for oslo.middleware to new release 4.0.1 https://review.opendev.org/709740 | 14:32 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: update constraint for oslo.vmware to new release 3.1.0 https://review.opendev.org/709741 | 14:35 |
| *** iurygregory has joined #openstack-requirements | 15:05 | |
| *** irclogbot_2 has quit IRC | 15:27 | |
| *** irclogbot_0 has joined #openstack-requirements | 15:29 | |
| *** e0ne has joined #openstack-requirements | 15:49 | |
| *** e0ne has quit IRC | 15:54 | |
| *** e0ne has joined #openstack-requirements | 16:09 | |
| *** e0ne has quit IRC | 16:16 | |
| prometheanfire | hmm, reqs being generated don't have python_version correct | 16:29 |
| prometheanfire | smcginnis: how does the releases repo submit jobs to reqs? (doesn't seem to obey python_version) | 16:30 |
| smcginnis | prometheanfire: Do you have an example? | 16:31 |
| smcginnis | It's just a script in the release playbooks. | 16:31 |
| smcginnis | Oh, I see https://review.opendev.org/#/c/709740/1/upper-constraints.txt | 16:32 |
| smcginnis | Odd. I thought that had been working right. I wonder if something changed? | 16:32 |
| prometheanfire | smcginnis: ya, I agree with you, though I will say that the reason I want them gone from reqs/constraints is because the service as lib thing | 16:32 |
| smcginnis | In a meeting now, but I will take a look a little later. | 16:33 |
| * prometheanfire shrugs | 16:34 | |
| *** camelCaser has quit IRC | 16:42 | |
| *** camelCaser has joined #openstack-requirements | 16:43 | |
| smcginnis | https://opendev.org/openstack/project-config/src/branch/master/roles/copy-release-tools-scripts/files/release-tools/update_constraints.sh | 16:44 |
| prometheanfire | fun | 16:45 |
| smcginnis | I think the issue is this? https://opendev.org/openstack/project-config/src/branch/master/roles/copy-release-tools-scripts/files/release-tools/update_constraints.sh#L107 | 16:45 |
| *** iurygregory has quit IRC | 16:48 | |
| openstackgerrit | Merged openstack/requirements stable/rocky: update constraint for keystonemiddleware to new release 5.2.2 https://review.opendev.org/709595 | 16:50 |
| openstackgerrit | Merged openstack/requirements stable/rocky: update constraint for keystoneauth1 to new release 3.10.1 https://review.opendev.org/709596 | 16:50 |
| openstackgerrit | Merged openstack/requirements stable/train: update constraint for keystoneauth1 to new release 3.17.2 https://review.opendev.org/709632 | 16:50 |
| *** udesale has quit IRC | 17:04 | |
| *** vishalmanchanda has quit IRC | 17:11 | |
| *** e0ne has joined #openstack-requirements | 17:16 | |
| *** evrardjp has quit IRC | 17:39 | |
| *** evrardjp has joined #openstack-requirements | 17:39 | |
| *** e0ne has quit IRC | 18:27 | |
| *** e0ne has joined #openstack-requirements | 18:28 | |
| smcginnis | prometheanfire: Think I have a fix - https://review.opendev.org/#/c/709809/ | 18:30 |
| *** hberaud has quit IRC | 18:40 | |
| *** ralonsoh has quit IRC | 18:42 | |
| *** e0ne has quit IRC | 18:44 | |
| prometheanfire | cool | 18:46 |
| *** e0ne has joined #openstack-requirements | 22:20 | |
| *** e0ne has quit IRC | 22:21 | |
| *** e0ne has joined #openstack-requirements | 22:21 | |
| *** migawa|lunch is now known as masayukig | 22:39 | |
| *** e0ne has quit IRC | 23:34 | |
| *** tosky has quit IRC | 23:52 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!