Friday, 2022-02-25

opendevreviewYi Feng proposed openstack/keystoneauth master: [WIP] OAuth2.0 Client Credentials Grant Flow Support  https://review.opendev.org/c/openstack/keystoneauth/+/83073401:47
slaweqgtema dtantsur stephenfin hi, can You check https://review.opendev.org/c/openstack/python-openstackclient/+/830289 when You will have few minutes? Thx in advance12:21
opendevreviewMerged openstack/python-openstackclient master: Add support for setting extra DHCP options on existing ports  https://review.opendev.org/c/openstack/python-openstackclient/+/83028913:26
*** Guest229 is now known as diablo_rojo_phone14:01
hberaudHello, please can someone of your team have a look to https://review.opendev.org/c/openstack/releases/+/82908114:16
elodilleshi sdk team! can you please double check the release patch for openstacksdk (and my comment on it?): https://review.opendev.org/c/openstack/releases/+/82908114:16
elodillesthanks! :)14:16
elodillesjinx :)14:16
hberaud:)14:17
gtemaI have pretty clear expressed why it can't be released now14:17
elodilles"no changes can be release at the moment. Yoga will will stay at current version"14:19
elodillesyou mean this? ^^^14:19
elodillesanyway, thanks for the response, just wanted to double check14:20
fungii guess that means the project's development schedule is no longer aligned with the coordinated openstack release, and it should be switched to the independent release model?14:22
fungithat way it can release (or not) whenever it likes without needing to follow schedules set by the openstack release managers14:23
gtemano, there were releases in this cycle14:23
gtemarelease model of sdk was always in reality a question, but back those days decided to stick to that, cause sdk is used by i.e. nova14:24
fungibut it's not observing the release freeze as there's new work going into master14:24
gtemathere were no features added required for the depending components14:25
gtemathus from my pov latest release can be used as is without any issues14:25
gtemachanges from feature branch landed quite some time ago and issues blocking release has been identified14:27
fungii understand, it's more a question of it currently being marked as using a different set of rules. this is the release model the release managers think openstacksdk is following: https://releases.openstack.org/reference/release_models.html#cycle-with-intermediary14:28
fungi"...commit to produce a release near the end of the 6-month development cycle to be used with projects using the other cycle-based release models that are required to produce a release at that time"14:29
gtemaI know. I fear there is currently no other suitable model for sdk as being that even it is something "special". In any way this is currently an exception, rather then the rule14:32
gtemaa big batch of changes were done in the feature branch and it got merged during suitable time frame14:33
fungiwell, it's not all that dissimilar from some other projects which follow the independent release model14:33
gtemasadly issues requiring further work were also identified (i.e. hurting Zuul) and as such release need to be postponed unless those got worked on14:34
gtemayeah, but if I read descriptions correctly just because being now considered a lib of some core components it must follow this model14:34
fungithere are lots of lib dependencies which aren't managed as part of the coordinated release. in fact, most of openstack's dependencies aren't even developed within openstack at all14:35
gtemawell, if tc/whoever else is interested are not really objecting that for SDK - I have nothing agains14:36
gtemaI just recall there were some discussions on that and it was decided to leave it like it is now14:37
fungino worries, just suggesting that as it might create lower levels of friction since the project is doing development at a point in the cycle where libs are supposed to be frozen until the stable branches are cut14:40
fungibut the release team can presumably branch stable/yoga from the 0.61.0 tag back in december14:41
gtemasure. As mentioned, I was just somewhy keeping in the back of my mind there were real reasons for this decision14:42
gtemahm, stable branches of SDK is another dirty story14:42
fungiright, independent release model doesn't require stable branches either14:46
fungi(allows them as needed, but does not require them cycle after cycle)14:47
gtemamaybe even that was somehow a reason for keeping sdk on this model14:48
gtemadecoupling sdk from releases is easy and makes sense on one side, but it will definitely face some resistance on release packaging side 14:49
fungimaybe, though if you follow semver conventions fairly closely and increment the minor revision on any dependency changes, you can still safely add a stable branch and tag patch level revisions on it after the fact14:49
fungiproblem arises if you change requirements.txt in 0.61.1 and then want to branch from 0.60.0 later, there's no room to tag stable point releases14:50
gtemaokay, would need to consider this again and potentially discuss during ptg14:51
fungiyeah, it would certainly be a good conversation to have14:51
diablo_rojo_phone+214:52
fungier, my example was slightly off. i meant to say "...if you change requirements.txt in 0.61.1 and then want to branch from 0.61.0 later..."14:56
opendevreviewRajat Dhasmana proposed openstack/python-openstackclient master: compute: Add support for microversion 2.91  https://review.opendev.org/c/openstack/python-openstackclient/+/83101417:15
opendevreviewMerged openstack/keystoneauth master: Fix bindep for current rpm based distributions  https://review.opendev.org/c/openstack/keystoneauth/+/83021118:22
opendevreviewMerged openstack/keystoneauth master: setup.cfg: Replace dashes by underscores  https://review.opendev.org/c/openstack/keystoneauth/+/82799318:25

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!