| opendevreview | OpenStack Proposal Bot proposed openstack/requirements master: Updated from generate-constraints https://review.opendev.org/c/openstack/requirements/+/845871 | 02:46 |
|---|---|---|
| opendevreview | Matthew Thode proposed openstack/requirements master: Updated from generate-constraints https://review.opendev.org/c/openstack/requirements/+/845871 | 03:03 |
| zigo | sean-k-mooney: I uploaded to Experimental especially to find the issues in 4.6.0 with the pseudo-excuse page, though Unstable still has 3.2.0. | 06:16 |
| zigo | The thing is other software in Debian need >= 4.4.0, which is why a few DDs are pushing hard so I update jsonschema to something more recent. | 06:16 |
| zigo | Which is why I'm asking if something can be done. | 06:16 |
| zigo | 3.2.0 from nov 2019 is quite old ... :P | 06:17 |
| rpittau | hello everyone! We saw requirements dropped the u-c value for coverage for python3.6 recently, that broke our sushy-tools python3.6 tox job (we're keeping compatibility for the time being as it's branchless), any suggestion on how to work around that/fix it? | 08:04 |
| sean-k-mooney | zigo: right but as i noted on the mailing list nova is not way its on 3.2.0 currently | 08:09 |
| sean-k-mooney | zigo: its on 3.2.0 becaus of warlock which is used by glance cleint | 08:10 |
| zigo | Right. | 08:10 |
| zigo | Thanks for this investigation. | 08:10 |
| sean-k-mooney | zigo: warlock is the lib with the <4 cap so i wonder how hard it woudl be to drop that | 08:10 |
| zigo | Hopefully, the fixed warlock can be released. | 08:10 |
| zigo | I wouldn't mind dropping warlock as you suggested... | 08:11 |
| sean-k-mooney | they have not merged any patches in 2 years | 08:11 |
| zigo | One less thing to maintain is always good. | 08:11 |
| sean-k-mooney | im thinking the project is pretty dead | 08:11 |
| sean-k-mooney | so ya i think we shoudl drop it if we can | 08:11 |
| zigo | It used to be maintained by bcwaldon, though I'm not sure if he's still in the OpenStack game. | 08:12 |
| zigo | (he was a Glance maintainer) | 08:12 |
| sean-k-mooney | ya is galnce client still supported by the way | 08:15 |
| sean-k-mooney | as in has everything been ported to osc | 08:15 |
| zigo | Looks like Brian is now working at: https://antaris.space/ so not OpenStack related anymore ... | 08:15 |
| zigo | I wouldn't mind if we dropped glanceclient ! :) | 08:16 |
| sean-k-mooney | novaclint is partly deprecated by the way. its supported as a lib going forward but the cli part is frozen | 08:16 |
| sean-k-mooney | eventually we want to drop it and have people use sdk but no timeline for that | 08:17 |
| zigo | However, if we can patch the issue out of glanceclient, that'd be better, so I could fix it in Debian Unstable (before Zed is released). | 08:17 |
| sean-k-mooney | i know neutron is in a similar trasition not sure about glance | 08:17 |
| sean-k-mooney | ya so that woudl be my path forward | 08:18 |
| sean-k-mooney | patch the issue out fo glance and drop warlock form global-requiremetns | 08:18 |
| sean-k-mooney | then update json scep to say 4.4 if it does not rbeak anything then fix what is left to get 4.6 working in nova and then update uc to 4.6 | 08:19 |
| frickler | zigo: sean-k-mooney: quite a can of worms we have there | 08:43 |
| frickler | heat-translator 2.5.0 requires python-glanceclient, python-glanceclient 4.0.0 requires warlock, warlock 1.3.3 requires jsonschema<4,>=0.7, but you have jsonschema 4.6.0 which is incompatible. | 08:44 |
| zigo | Agreed. Though we can't stay stuck on a 3 years old version of jsonschema, at some point it needs to be fixed. | 08:44 |
| frickler | talk to glance, I'd say. not sure what warlock actually does | 08:44 |
| frickler | we also have some other issues. like anyjson is no longer buildable. our whole CI only works because we are using a historically built wheel for it | 08:45 |
| frickler | things will fail miserably once we try to run u-c updates on python3.10 | 08:45 |
| sean-k-mooney | frickler: i commented out anyjson to get to the warlock issue | 08:46 |
| frickler | ah, now I see that you found warlock already in the backlock | 08:46 |
| frickler | sean-k-mooney: I used our wheels cache instead, but yes | 08:47 |
| frickler | python-nss also seems unbuildable with openssl3 on Jammy | 08:47 |
| sean-k-mooney | so what would it take to drop both fo those by the way | 08:47 |
| sean-k-mooney | warlock looks simple to drop | 08:47 |
| sean-k-mooney | is anyjson used much | 08:47 |
| frickler | not so much I'd say, at least within openstack proper https://codesearch.openstack.org/?q=anyjson&i=nope&literal=nope&files=&excludeFiles=&repos= | 08:50 |
| frickler | not sure how active murano-agent is these days | 08:50 |
| frickler | rpittau: use yoga constraints for that job? | 08:51 |
| rpittau | frickler: yep, that was taken into consideration, just maybe wondering if there was an alternative | 08:54 |
| frickler | rpittau: the only alternative I see would be to stop using global-requirements for that project and set your own instead | 09:00 |
| rpittau | frickler: thanks! I saw that done by swift for example, we'll discuss if it's worth keeping py3.6 support at this point | 09:01 |
| *** tosky_ is now known as tosky | 09:42 | |
| sean-k-mooney | was it? | 11:01 |
| sean-k-mooney | that seam like not a good thing in general | 11:02 |
| sean-k-mooney | e.g. if some project dont use global-requirements anymore | 11:02 |
| sean-k-mooney | dont we still requrire co installablity | 11:02 |
| sean-k-mooney | e.g. for it to be an offical project in the openstack namespace | 11:02 |
| sean-k-mooney | from a governace perspecitve i tought the PTI with global requirements was a requirement | 11:03 |
| sean-k-mooney | frickler: so lookign at the second wip patch for jasonschema 4.6.0 vs my 4.0.1 a similar number of project are failing on both | 11:05 |
| sean-k-mooney | so we proably shoudl go strait to 4.6.0 | 11:05 |
| sean-k-mooney | and just have a one time pain | 11:05 |
| sean-k-mooney | i was hoping we could advance the bump a few times in the 4 seriese | 11:05 |
| sean-k-mooney | to reduce the scope of what we needed to fix | 11:06 |
| sean-k-mooney | which is why i started with 4.0.1 but if that fails and 4.6.0 fails there proably isnt much value in doing a seach to see if any of the intermedaries work | 11:06 |
| frickler | sean-k-mooney: ack, going to latest seems preferrable | 11:08 |
| sean-k-mooney | we should be able to use a depend on form nova to the requriement patch to develop a patch to fix the compatiblity yes | 11:09 |
| sean-k-mooney | and then flip that | 11:09 |
| sean-k-mooney | once it owrks | 11:09 |
| sean-k-mooney | well or just do it locally i guess | 11:10 |
| sean-k-mooney | we can manuyally install the newer version with pip | 11:10 |
| sean-k-mooney | and then fix what is broken and just have the bump patch depend on it | 11:10 |
| sean-k-mooney | gmann: you mentioned another lib | 11:10 |
| sean-k-mooney | move form v4 to v7 or something like that | 11:11 |
| sean-k-mooney | were you working on a patch for that or planning to | 11:11 |
| *** tosky_ is now known as tosky | 11:55 | |
| opendevreview | OpenStack Proposal Bot proposed openstack/requirements master: update constraint for taskflow to new release 5.0.0 https://review.opendev.org/c/openstack/requirements/+/845980 | 12:25 |
| frickler | sean-k-mooney: did you see this? openstack-placement 1.1.0 depends on jsonschema<3.0.0 and >=2.6.0 | 12:28 |
| frickler | ah, wait, that may be old versions due to pip trying to resolve other things | 12:29 |
| frickler | https://zuul.opendev.org/t/openstack/build/49a23a47b6cd460daf4a334917fec7c0 | 12:29 |
| sean-k-mooney | i didnt see that no but ill check placment | 12:31 |
| sean-k-mooney | https://github.com/openstack/placement/blob/master/requirements.txt#L14= | 12:32 |
| sean-k-mooney | jsonschema>=3.2.0 # MIT | 12:32 |
| sean-k-mooney | so we shoudl be fine assuming its compatible | 12:32 |
| frickler | sean-k-mooney: yes, I think pip output is misleading. iiuc the real conflict is with warlock/python-glanceclient, which nova also depends on | 12:37 |
| frickler | just designate looks like it will need a proper fix, too https://zuul.opendev.org/t/openstack/build/75b6d1dff31e4af8b239f7b2e3086bbc | 12:37 |
| sean-k-mooney | we should really move to sdk :) | 12:53 |
| *** tosky__ is now known as tosky | 13:17 | |
| *** tosky is now known as Guest2169 | 13:29 | |
| *** tosky__ is now known as tosky | 13:29 | |
| *** tosky__ is now known as tosky | 13:58 | |
| johnsom | frickler Have you started a patch for designate or is it something I should work on? | 14:33 |
| frickler | johnsom: I didn't start yet, feel free to have a look | 14:39 |
| johnsom | frickler Ack, I will look into it. | 14:39 |
| johnsom | Thanks | 14:39 |
| prometheanfire | just syncing now, looks like talk about removing caps only? | 15:18 |
| frickler | prometheanfire: for jsonschema, the cap is implicit via a glanceclient dep, no direct action needed on the requirements side | 15:42 |
| frickler | prometheanfire: for anyjson, it needs to be dropped completely IMO. the latest version is from 2012 | 15:43 |
| frickler | it's one the the things listed in https://review.opendev.org/c/openstack/requirements/+/813302/3/upper-constraints.txt | 15:44 |
| frickler | gotta refresh those test jobs and maybe try to run propose-updates on py3.10, too | 15:45 |
| frickler | that'll certainly bring out more cruft | 15:45 |
| prometheanfire | fair on the anyjson | 15:47 |
| prometheanfire | if we can get 3.10 that'd be nice, but atm we can't run all the pythons in one place, which limits us | 15:47 |
| prometheanfire | openstack/murano-agent is the only thing that really uses anyjson, I think | 15:48 |
| johnsom | FYI, I have posted the designate fix for jsonschema > 4 support: https://review.opendev.org/c/openstack/designate/+/846014 | 15:49 |
| opendevreview | Dr. Jens Harbott proposed openstack/requirements master: WIP: Run a job without the pre-built wheels https://review.opendev.org/c/openstack/requirements/+/813292 | 16:09 |
| opendevreview | Dr. Jens Harbott proposed openstack/requirements master: DNM: Show failed builds and dependents https://review.opendev.org/c/openstack/requirements/+/813302 | 16:17 |
| opendevreview | Merged openstack/requirements master: update constraint for taskflow to new release 5.0.0 https://review.opendev.org/c/openstack/requirements/+/845980 | 16:20 |
| gmann | sean-k-mooney: that is on jsonschema only if we move to new version of it then I can work on nova using the v7 validator from jsonschema. but moving to new jsonschema still have v4 draft which is used in nova so no backward incompatible change due to that. | 23:16 |
| gmann | sean-k-mooney: as summary, I can work on that part later once jsonschema version is bumped | 23:17 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!