Saturday, 2026-06-27

@fungicide:matrix.orghere's hoping the crawlers leave us alone00:23
@rm_work:matrix.orgIs there a way in the zuul.opendev.org browser to see all changes with a specific *tag* or do I need to just assemble a big list lol03:31
@rm_work:matrix.orgalso, zuul is smart enough to kill existing runs on old patchsets if I push a new patchset right?03:35
@rm_work:matrix.orgtrimming it down to run only the one devstack test I care about for now so I stop spamming a ton of jobs that don't matter T_T03:36
@rm_work:matrix.orgoh also -- no more x/ namespace for devstack plugins, but... I have *another one* I have been working on for the alternate oslo.messaging RPC system (HTTP-RPC) which seems to be literally textbook "should be a devstack plugin repo" I think? so I would have to propose that in the openstack/ namespace directly? :/ Clark 04:28
@rm_work:matrix.orgalso feel free to ignore me until after the weekend :P04:28
@rm_work:matrix.orgthis: https://review.opendev.org/c/openstack/oslo.messaging/+/99460104:58
@fungicide:matrix.orgrm_work: the zuul status dashboard just treats queue items as opaque identifiers, so it's not aware of things like gerrit hashtags (it may be aware of gerrit change topics in order to properly indicate dependency groups when gerrit's "merge whole topic" feature is enabled, but since we don't use that in opendev i don't know)10:21
@fungicide:matrix.orgit *is* smart enough to cancel jobs and dequeue the old patchset when you push a new one, yes (and if you have dependencies in changes you may see them instantly report a failure condition because their parent changed while jobs were in progress)10:22
@fungicide:matrix.organd yes, for changes under heavy experimentation, it's helpful to temporarily trim jobs out of the zuul configuration you're not focused on results from yet10:23
@fungicide:matrix.orgadding projects to the `openstack/` git namespace needs the approval of an official openstack project team willing to adopt it from a governance perspective, for example the qa or oslo team in that specific case10:25
@fungicide:matrix.orgbut you could also stick it in a new unofficial namespace like `devstack-plugins/` or `arglebargle/` (just not `x/`)10:26
@fungicide:matrix.orgprojects can also move from other namespaces into `openstack/` later if a project team agrees to adopt them officially at some point in the future, we're doing that in a couple of weeks because `x/cursive` is getting adopted by openstack oslo and becoming `openstack/cursive`10:29
@harbott.osism.tech:regio.chathttps://review.opendev.org/c/openstack/oslo.messaging/+/994601 looks good to me as it is, unless a different project also needs the same functionality? maybe it could even integrate https://opendev.org/openstack/devstack-plugin-kafka and get rid of that separate repo? more of a discussion for the oslo team, though16:47
@rm_work:matrix.orgyeah I mean I guess it COULD stay as an oslo.messaging in-tree plugin but... likely it would make more sense as a split out plugin, so I will try adding `devstack-plugins/devstack-plugin-http-rpc` as fungi suggested and when it's all settled we can move it to the openstack namespace20:53
@clarkb:matrix.orgplease no20:55
@clarkb:matrix.orgI really want to get away from creatnig unncessary work on the opendev team20:56
@clarkb:matrix.orgrenaming repos requires a gerrit downtime too so is disruptive to everyone20:56
@clarkb:matrix.orgwe shouldn't create repos with the plan to renae them later20:56
@rm_work:matrix.orghmmm ok20:56
@clarkb:matrix.orgjust create them where they should go and leave them there20:56
@rm_work:matrix.orgI mean I am open to whatever20:56
@rm_work:matrix.orgwell, it would... be in openstack/20:56
@rm_work:matrix.orgbut the change it's for isn't even merged yet20:57
@rm_work:matrix.orgso it seemed like that'd be hard to sell20:57
@rm_work:matrix.orgyou can see that CR is the end of a giant stack20:57
@rm_work:matrix.organd that stack has been open for A WHILE though I am hoping we can make progress towards merging it now that it is a multi-org investment20:57
@clarkb:matrix.orgright thats one of the great things about gerrit and zuul. You can do all of this speculatively without committing to much20:59
@rm_work:matrix.orgI literally do not care and just want to do the right thing, hopefully that is clear :D 20:59
@rm_work:matrix.orgI just need guidance on "what is the right thing" lol20:59
@clarkb:matrix.orgyou can show everything wroks end to end across arbitrary code bases. Etc then if you need to createa a repo its an easy thing to decide on21:00
@clarkb:matrix.orgyou'll have to ask openstack maintaniers where they want code to live. All I'm saying is if you paln to create a repo just to rename it later you are creating unnecessary work for me and forcing a gerritdowntime on everyone and weshouldnt' do that21:00
@rm_work:matrix.orgwell, I can't CREATE A REPO speculatively right? so basically just keep it in-tree (where it doesn't really belong) for now and point to it as proof stuff works, get the patches it is for merged, THEN make the repo and move the code where it belongs?21:00
@rm_work:matrix.orgever so slightly annoying just because it feels gross for the time being, but i think it doesn't pose any real technical problems, so that's fine. I'll try to be patient :D21:01
@clarkb:matrix.orgI would argue it is the opposite of gross you just have biases from using other tools21:01
@clarkb:matrix.orgGerrit is built to accept code proposals against refs in temporary magical changes refs21:01
@rm_work:matrix.orgheh I mean I complain about other tools all day long because I have biases from using Gerrit for too long :P21:02
@clarkb:matrix.orgit isn't gross to use Gerrit this way21:02
@rm_work:matrix.orgthe issue here is that I have code where it doesn't belong, temporarily21:02
@rm_work:matrix.orgI love the speculative stuff21:02
@rm_work:matrix.orgI use Depends-On heavily here21:02
@rm_work:matrix.orgit is A+++++++21:02
@clarkb:matrix.orgthe idea that we need to create a new repo for every new idea is very much a github repos are cheap style workflow. And its fine there. Gerrit is different and using cheap refs instead of repos is fine with Gerrit21:03
@rm_work:matrix.orgwell it's more like "that really should be the final home"21:03
@clarkb:matrix.orgbut also if you bring it up with opesntack maybe they say go ahead and make a repo and then we don't have to rename anything and you can just use that21:03
@rm_work:matrix.orgliterally the canonical example of "when to create a new devstack plugin repo" is "new RPC transport"21:03
@rm_work:matrix.org(per docs)21:03
@clarkb:matrix.orggreat then it should just go into openstack imo if the plan is to have openstack maintain it21:03
@clarkb:matrix.orgfrom day one I mean. Don't plan to rename it later21:04
@rm_work:matrix.orgyeah the issue is just that it's a speculative creation of a repo21:04
@rm_work:matrix.orgso again, I will just keep it in the arbitrary code storage area of oslo.messaging even though it doesn't actually belong there21:04
@rm_work:matrix.orgit is functionally fine21:04
@rm_work:matrix.orgdo appreciate the advice :D21:05
@rm_work:matrix.orgalso, would love anyone's opinion on this ML draft if you have a minute, not really part of your focus here, I just trust your opinions :P21:06
https://paste.opendev.org/raw/bruwFXVlu01EdnvJP2rN/
@rm_work:matrix.org(my other emerging effort, not related to the RPC thing)21:06
@rm_work:matrix.orgI've now got everything passing devstack-tempest served by Hypercorn ASGI except glance/neutron but I expect to figure those out soon... Phase 0 basically complete21:07
@rm_work:matrix.orgI should probably link https://review.opendev.org/q/hashtag:%22asgi%22+(status:open%20OR%20status:merged) in there21:08

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!