Friday, 2026-05-08

mnasiadkanoonedeadpunk: are you ok with gouthamr’s response in https://review.opendev.org/c/openstack/project-config/+/959904? Any team can override toggleWipState locally (in their acts) - so I think I’m leaning towards merging it05:21
mnasiadkaIn their acls, bloody correction mechanism :)05:22
opendevreviewMerged openstack/openstack-manuals master: [www] Set 2024.2 Dalmatian as End of Life  https://review.opendev.org/c/openstack/openstack-manuals/+/98715006:06
noonedeadpunkmnasiadka: so override in order to allow or prohibit?07:05
mnasiadkanoonedeadpunk: prohibit (or rather limit to a -core group or something like that)07:05
noonedeadpunkAs I am not sure what consistency are we talking there - I can not unset as a core somebodey elses -W, can I07:05
noonedeadpunkSo consistent behavior is actually to keep current default...07:06
noonedeadpunkbut I guess whatever07:06
noonedeadpunkIf only I see it this way, I am fine to agree to disagree07:06
mnasiadkaCurrent default is a bit irritating to be frank, so either we go through all repos and grant -core group toggling WIP state (which is not the nicest job around) or we go with gouthamr’s proposal (and limit on per group if needed)07:07
noonedeadpunkYeah, this is smth I get. And I already changed ACLs for OSA :D07:08
noonedeadpunkBut if we're talking about consistency...07:08
noonedeadpunkLike the one thing I am kinda "afraid" of - I know bunch of people that are really humble, and not ready to show up result of their work07:09
noonedeadpunkSo it is really hard to convince them to push changes to gerrit, because they are gonna be judged and reviewed07:09
noonedeadpunkAnd WIP is kinda a "safe harbour" for them07:10
mnasiadkaWell, you can rebase or edit to get rid of workflow -1, wipstate is different - now in majority of repos only the owner can change it :)07:10
noonedeadpunkRight, rebase or edit indeed does work...07:10
noonedeadpunkWell, as I said, I am fine agree to disagree here I guess07:10
mnasiadkaAnd around these humble people - if only gerritbot would not post URLs to patches that are WIP - I would agree :)07:10
mnasiadkaSo maybe let’s go with that and see if there will be any abusers of this - we can always go back07:11
mnasiadkaAnd that’s in line with other comments on that patch07:13
noonedeadpunksure07:40
noonedeadpunkriiight... I really wonder if we actually should patch gerritbot...07:41
noonedeadpunkas also WIP patches in IRC are not very meaningful, imo07:41
noonedeadpunkbut that is completely different topic07:42
fricklerwell I often do look at wip patches when I notice them on IRC and even comment on them, so I wouldn't necessarily agree on this07:53
gouthamrif you’re thinking about the way gerrit is writing their APIs, the use case you have doesn’t make sense, noonedeadpunk .. you as core wanting to remove someone’s W-1..08:14
gouthamryou can’t overwrite a label added by someone else; that would get complicated, and abused.08:14
noonedeadpunkoh, sure, I am not arguing that removing label makes sense08:15
gouthamrty for discussing this btw and driving a resolution, mnasiadka08:15
noonedeadpunkbut I am not sure that WIP should be treated differently from regular labels08:16
noonedeadpunkAnd then why not to disable WIP as a whole and keep using -W08:16
gouthamrI don’t know if that’s an option?08:16
gouthamrI don’t like it myself.. it’s easy to get yourself into a WIP without intending to08:17
noonedeadpunkWell. Given that gerrit is used not only for openstack... probably not08:17
noonedeadpunkI don't know tbh08:17
noonedeadpunkMaybe we can get an ACL or smth disabling/prohibiting it for all openstacvk projects08:18
mnasiadkaIn an ideal world WIP is nice, because you can mark your patch as being WIP - as long as you’re active08:21
mnasiadkaBut in cases when somebody picks up your patch after N months of inactivity on it, that process gets complicated08:21
gouthamr+108:23
noonedeadpunkthe problem is that we can't drop -W and leave just positive option anyway...08:26
noonedeadpunkSo we have 2 things serving around same purpose08:26
mnasiadkaW-1 for me says do not merge this, and is used in other cases than pure WIP08:27
noonedeadpunkW-1 is most useful to abort gate jobs for me though. But I am not sure what -W is not covering that pure WIP does08:31
kinruijust a reminder, the workflow -1 label was something we added as a workaround when we moved our gerrit fork that had a wip flag, with the idea that we could get rid of workflow -1 eventually if gerrit added a wip feature upstream11:38
kinruier, when we moved *off* our gerrit fork11:39
sean-k-mooneykinrui: whil tha tis true to do that properly we need to modify the acls on every repo so that core team meber can remove the wip flag16:57
sean-k-mooneyalso by defualt gerrit filters out wip patches in some dashboard16:58
clarkbI think part of the problem is the overuse of WIP16:58
sean-k-mooneyi.e. if you just go to changes adn click on open it redirects to https://review.opendev.org/q/status:open+-is:wip16:58
clarkboften times it feels like it is used when the author isn't confident in the change and not when the author is trying to avoid feedback. But really the way gerrit handles it the idea is "avoid bothering people/no feedback and prevent merging"16:59
clarkbwhat the label addresses is prevent merging without the second bit16:59
sean-k-mooneyright i think the "this is not ready to merge but i woudl like early feedback" is the way we use it most often17:00
sean-k-mooneytypiclly i use [WIP] in the commit to also delinate this17:00
sean-k-mooneybecause often thirdpary ci system will not run if you have -W or wip set 17:01
sean-k-mooneybut i might want that feedback17:01
sean-k-mooneyas a core reviewr i generally want ot see wip patches by default and i can do that by adjucting hte filter17:01
sean-k-mooneybut what i cant do today without modifying the acls is take over a wip patch form a contibutor that started the work17:02
sean-k-mooneyi.e. they pushed it with WIP then lsot interst and i could not remove it17:02
sean-k-mooneyso today i fine the native feature harmful until we correct that17:03
sean-k-mooneynoonedeadpunk: i would not like use to patch gerrit bot to avoid posting wip patches to irc17:03
sean-k-mooneyi do want to see those so i cna give basic directional input on the patch if tis in an area that i care about17:04
sean-k-mooneyim not really agaisnt changign the text of -W as shown in the ui to encuage it to be used only for "do not merge this"17:06
sean-k-mooneybut we still need +W for ok to merge17:07
sean-k-mooneyso the lable cant go away entirly17:07
clarkbat the same time it is trivial to repush a change. And typically when you do that for a wip change you don't lose much of value except for waiting on ci results (because typically there hasn't been a ton of review at that point)17:07
clarkbyes I like the intentional act of "please merge this" being independent of the code review state17:08
sean-k-mooneyclarkb: the last tiem i tried that the wip flag was not cleared automaticly17:08
clarkbsean-k-mooney: you have to push a new change17:08
sean-k-mooneywell atleast it was not cleard when i did a rebase of the patch via the ui17:08
clarkbthen abandon the WIP change17:08
sean-k-mooneyi didnt try doing it via git-review17:08
sean-k-mooneyoh17:08
clarkbrebasing does not make a new change it merely updates the existing change with a new patchset that has a new parent17:08
sean-k-mooneywll the issue with that is you lose the orginal author17:08
sean-k-mooneymaybe that is not true17:09
clarkbyou shouldn't that will be preserved by git17:09
clarkbthe gerrit change owner will change but the author shouldn't17:09
sean-k-mooneyright so i woudl need to change the change id in the commit message which will change the commiter17:09
sean-k-mooneybu tyour right the author shoudl be perserved17:10
clarkbyes but not the author17:10
sean-k-mooneyya its a bit of a pain but workable17:10
*** kinrui is now known as fungi22:44

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