[10:43:09] tgr, addshore: hey! shall we poke at the remaining things on https://gerrit.wikimedia.org/r/c/mediawiki/core/+/440176 ? [10:43:50] there are a few things that seem to still be untested. [10:44:17] and a few tests failed. we could checkout master and try that again, to see if it makes any difference [10:44:32] tgr: fyi https://phabricator.wikimedia.org/T196585#4281359 [10:47:52] DanielK_WMDE: yeah, I'll check with master [10:51:49] DanielK_WMDE: 'Provide backwards compatibility for RevisionInsertComplete hook. [10:52:05] ... was not applied to the test branch on mcr-full, right? [10:52:19] no. fetch & rebase would fix that [10:52:29] may or may not conflict... [10:52:45] or try a cherry-pick. whichever [10:52:57] strangely, revision-create on EventBus did work [10:53:25] tgr: are you going to switch mcr-full to master? then soon? then I won't do any testing tehre until you tell me it's back on the StorageLayerTesting branch [10:53:39] yeah, I'm switching now [11:06:23] o/ [11:06:33] just got out of a meeting, but going back into another one in 9 mins [11:07:42] tgr: DanielK_WMDE feel free to merge things into core master if and when they are ready [11:07:55] I can help testing again at some point in ~1 hour or less [11:16:37] DanielK_WMDE: the good news is, all errors are present on master as well [11:16:59] Echo was just me messing up; ConfirmEdit might just be misconfigured [11:17:14] Flow seems pretty broken: http://dev-mcr-full.wmflabs.org/w/index.php?title=User_talk:Tgr-test-1&action=history [11:27:14] ConfirmEdit needs a scritp to run to pre-generate the captchas, iirc [11:27:27] i'll look into flow in a bit [11:27:35] have to follow up on some techcom action items first [11:28:23] tgr: ...is that flow error with master, or only on our branch? [11:28:34] master, too [11:28:41] ah. [11:28:42] *sigh* [11:28:44] none of the errors are specific to the branch [11:28:59] ok, good to know. can you say so on the wiki page? [11:29:38] I'm guessing it uses parsoid to create a revision, parsoid is dead, and it does not handle that gracefully and ends up with an invalid revision somehow [11:30:09] and the parsoid issue is probably T166953 [11:30:12] T166953: npm::install: Maximum call stack size exceeded - https://phabricator.wikimedia.org/T166953 [11:30:21] trying to fix that now [11:39:51] ...and I actually fixed it half a year ago. I need to get into the habit of self-merging my vagrant patches. [11:55:15] *sigh* [12:06:51] ok, restbase & parsoid is now up and running. VisualEditor still not working. [12:07:11] nor is Special:EnableStrucutredDiscussions: Flow\Import\PageImportState::rollback: Expected mass rollback of all peer transactions (DBO_TRX set) [12:07:17] (this is still on master) [12:35:08] tgr: is it really worth spending time on this? [12:35:43] is VE active on beta? can we test there after merging? [12:36:25] this seems like a genuine Flow bug, but fair enough [12:37:00] tgr: yea... perhaps tell them :) but if we want to get the patches onto testwiki, we need to merge them today, and test them again on beta... [12:37:21] the PageUpdater patch is ready for merging, I haven't followed the other two closely [12:37:30] should I merge those too? [12:37:52] (T197234) [12:37:53] T197234: Expected mass rollback of all peer transactions (DBO_TRX set) - https://phabricator.wikimedia.org/T197234 [12:40:46] DanielK_WMDE: last plan is to merge into master, right? I'll proceed with PageUpdater then [12:42:41] tgr: yes [12:43:08] tgr: the RevisionStore patches got another round of reviews from addshore. [12:43:18] he didn't review again after i followed up [12:43:33] i hope he can do that as soon as he comes back from lunch [12:43:35] I can do, just eating lunch :) [12:45:36] addshore: VE doesn't want to run on mcr-full, and Flow is havign issues, too. We should be able to test both on beta, though [12:47:13] I'll fix VE, we'll want to use this later, and if the bug is in the vagrant role, it should be fixed anyway [12:47:23] should not block merging the patches though [12:48:08] tgr: thanks! [12:48:29] right, food done [12:48:31] review time [12:48:33] i'm wondering when we want to send a heads-up. when this lands in master (so, now)? or when it goes out with the train? [12:48:45] DanielK_WMDE: when it lands on master [12:49:14] to releng and... cloud? ops?... [12:49:22] i guess we want to explain what is happening, no train next week and if all goes accoridng to plan it will get the train the following week [12:49:30] ops and wikitech (i asked greg last night) [12:49:38] ok [12:49:59] can link to https://phabricator.wikimedia.org/T196585 for the funky branch stuff we are doing to get it on group0 for a whole week too [12:50:09] we could put a note into the deployment schedule for the next train. "beware that you are rolling out this thing". [12:50:31] i'm a bit confused what we need the extra branch for... [12:50:57] what not just do a "short train", i.e. regular deployment, but only group0? [12:51:02] or is that more confusing? [12:51:09] * addshore will leave that up to releng [12:51:16] fine :) [12:51:29] there will be a branch whatever happens, but as it is being cut tonight, maybe it doesn't want to follow the standard names [12:51:48] do we want to link to our test wiki in the mail? or rather aqwvoid that and ask people to test on beta (and on testwiki)? [12:51:49] also, that is a decision to make, is it a branch cut, or .8 with the 4 patches cherry picked onto it [12:52:10] I think getting testing on beta and test will make more sense than on our test wiki [12:52:38] yea... easpecially since i want to switch the mcr-full to MIGRATION_WRITE_BOTH asap. [12:52:50] then test all the stuff, then try schema migration. [12:52:56] ...and then MIGRATION_NEW [13:00:29] DanielK_WMDE: I think I'm happy with the revisionstore patches, i spent a very loong time staring at them last time I reviewed them and not much has changed [13:02:32] how do you remove the WIP label from a patch? [13:02:39] not that it matters much [13:03:22] the new automatic one? [13:03:35] i think by voting [13:05:24] DanielK_WMDE: one manual rebase of https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/416465/ needed [13:05:49] it's not automatic, there's a mark as WIP option in the menu [13:05:58] but no mark as not WIP [13:06:16] maybe it depends on ownership somehow [13:08:35] it has "mark as reviewd"... [13:08:40] not sure what that means exactly [13:09:30] yea, it's a bit strange [13:10:54] tgr: https://gerrit-review.googlesource.com/Documentation/intro-user.html#wip [13:11:26] "Alternatively, click Ready from the Change screen." [13:11:31] I don't see "ready" [13:12:22] oh, WIP mode suppresses notifications [13:12:23] yeah, me neither [13:12:27] and all my patches are WIP now? [13:12:32] this is a bit unfortunate [13:12:37] wtf? [13:13:00] I do get email notifications though [13:14:45] OTOH the part about not appearing on gerrit dashboards seems to be true [13:15:27] yea... that's unexpected. i'm not seeing some of the changes I really want to be seeing on my dashboard [13:16:17] paladox: any idea how marking patches as not WIP works? [13:16:22] I cant see how to remove it either [13:16:24] hmm [13:16:51] this should work: git push origin HEAD:refs/for/master%ready [13:16:57] oh i got it [13:17:03] I have a button called "start review" [13:17:22] perhaps only I see it as the change is originally mine [13:17:34] i don't see that on mine. [13:17:43] instead, i see "mark reviewed". as if the review was already started. [13:17:51] this is confusing [13:17:53] which change? [13:18:03] if I manually set something as WIP, I can't unset that either [13:18:09] oh [13:18:13] so maybe something is misconfigured [13:18:18] then maybe thats something to do with my gerrit access level? [13:18:29] it just seems broken :) [13:18:34] though I thought the only thing I could do was make repos :P [13:20:43] uhh it seems like the WIP flag stops zuul from merging it, too? [13:20:52] >.> [13:20:56] * addshore looks at the change [13:21:45] hmm, well, gerrit says "Cannot Merge" [13:22:32] I rebased it, lets try again? [13:23:17] gate-submit is running again, PS99! [13:24:57] tgr: do you use the old or new gerrit UI? [13:25:08] both [13:25:30] haha, okay, @DanielK_WMDE that could eb why you dont see the button i tked about though, as I am on the old UI still [13:25:48] the patch still says work in progress :( [13:26:08] I rebased the patch 40 minutes ago; the last core commit was two hours ago [13:26:16] it should not need rebasing [13:26:43] I'm confused now [13:26:48] do i need to rebase something? [13:27:14] I dont know D: [13:27:29] well, you need to rebase the 2nd revstore patch, but that doesnt relate to the WIP issues ;) [13:27:49] maybe we should go and wave our arms around toward some gerrit admins [13:28:33] DanielK_WMDE: so you have no start review button? :( [13:28:55] https://usercontent.irccloud-cdn.com/file/PnT96yI4/image.png [13:29:09] i have a "mark reviewd" button!" [13:29:10] I checked both interfaces, no button [13:29:21] DanielK_WMDE: hit the mark reviewed one and see what happens? [13:29:27] mark reviwed / mark unreviewed is a different thing [13:29:34] i did that for https://gerrit.wikimedia.org/r/c/mediawiki/core/+/406595/73 [13:29:36] and i gave a +1 [13:29:56] I have a WIP button on my non-WIP patches [13:30:03] now i have a "mark unreviewd" button :) [13:30:05] but I can't undo it if I click it [13:30:17] *sigh* [13:30:21] *sigh* [13:31:19] im poking people in releng [13:32:40] addshore: can you +2 this one? https://gerrit.wikimedia.org/r/c/mediawiki/core/+/406595/73 [13:33:06] DanielK_WMDE: it has a TODO of RELNOTES, but i guess we can followup with that? [13:33:59] mutante clicked the "set ready for review" button for us [13:34:03] addshore: we could. i'm honestly not sure these notes are needed [13:34:35] mutante is here in the office, i could grab him if we need hime [13:34:38] *him [13:34:44] lets see if it submits once the gate submit is done now [13:35:27] addshore: the first patch doesn't support any mopdes other than MIGRATION_OLD. so the release notes are not actually needed there [13:35:35] (I'm honestly not sure whether they are needed at all) [13:35:56] PageUpdater has merged :) [13:36:07] oh wait, it does support write-both [13:36:35] yup [13:36:55] but really - who are these release notes for? [13:37:23] anyway, if you want them, make a patch :) [13:38:52] * addshore doesnt care [13:38:53] ;) [13:39:51] DanielK_WMDE: mind rebasing https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/416465/ ? then the 3 are all in! [13:41:35] addshore: will do when the previous one has merged, to avoid confusion [13:44:07] addshore: hm, we should really remove the DNM markers before merging stuff ;) [13:47:09] * DanielK_WMDE is waiting for zuul [13:49:39] DanielK_WMDE: i DID [13:49:43] opps caps! :) [13:49:54] at least, from the first revstore patch [13:50:33] ah, good! [13:50:42] i just removed it from the second, locally [13:51:10] when zuuil is fone, i'll fetch, rebase, and push [13:59:26] Tgr not sure what you mean? You want to make a patch that does not have wip status? [13:59:43] Ah [13:59:45] I see now [13:59:57] In polygerrit there should be a start review button [14:00:21] In gwtui there should be a button for unmark it as wip but forgot what they call it [14:03:02] in poly gerrit it is this for me https://usercontent.irccloud-cdn.com/file/AhU0wV9i/image.png [14:03:21] but the button also appears in the old ui [14:03:53] paladox: what decides if a change is WIP in the first place? [14:04:03] I seem to have a very odd mix of things that appear in my WIp queue [14:04:13] If you do it through the ui [14:04:17] It’s a wip [14:04:26] Upstream are changing that [14:04:42] "do it through the UI"? [14:04:48] So that it’s preference based [14:05:08] Or you can do it through the command line [14:05:50] * addshore always submits his changes in exactly the same way [14:06:01] * addshore has never marked any as WIP in the UI or via cli [14:06:28] Heh [14:06:37] That’s a new feature in 2.15! [14:08:28] but then why are some of my changes already marked as WIP without my doing? :P [14:09:07] Ohhh [14:09:23] Did you do it through the command line? [14:09:31] nope [14:09:44] I just said, I have never marked anything as WIP in the cli or in the ui [14:10:12] i was wondering if there was some sort of commit message macthing, but that also wouldnt make any sense, as i have like 15 changes that say WIP that gerrit hasn't marked as wip [14:10:21] Ok [14:10:27] * paladox has no idea :) [14:10:35] But gerrit may have a parser [14:10:39] That does that [14:10:46] Just may not affect existing changes [14:11:43] DanielK_WMDE: patch #1 of rev store is in :) [14:17:49] addshore: sorry, got side tracked [14:17:51] will rebase now [14:17:56] surprisingly many conflicts [14:18:10] D: [14:18:25] im starting testing on beta and watching logs [14:29:38] addshore: new PS is up. https://gerrit.wikimedia.org/r/c/mediawiki/core/+/416465 [14:29:43] let's see what jenkins has to say [14:29:49] DanielK_WMDE: no changes? O_o [14:29:55] https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/416465/28..29 [14:30:05] what's the difference between "unreviewd" and "ready for review"? [14:31:13] uh. that's wrong. [14:31:17] there definitly are changes [14:31:23] that's a rebase, afterall! [14:31:40] maybe gerrit is trying to be "smart" about showing diffs for rebases? [14:32:26] gerrit should stop trying to be clever [14:32:37] but this is empty, too https://gerrit.wikimedia.org/r/c/mediawiki/core/+/416465/28..29/includes/Storage/RevisionStore.php [14:32:39] wtf? [14:33:01] are you sure you changed something :P [14:33:51] yes. line 453 to 456 of RevisionStore [14:34:02] merged in the RevisionInsertComplete stuff [14:34:20] aaaah, maybe it shouldnt show that [14:34:23] if that was all the changes [14:34:29] i dont know, silly gerrit [14:34:45] DanielK_WMDE: https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/416465/27..29 [14:35:31] addshore: https://gerrit.wikimedia.org/r/c/mediawiki/core/+/416465/28/includes/Storage/RevisionStore.php#450 vs https://gerrit.wikimedia.org/r/c/mediawiki/core/+/416465/29/includes/Storage/RevisionStore.php#451 [14:36:17] addshore: oh, you already put that in? [14:36:24] but why did that shouw up as a merge conflict again? [14:36:26] bah, whatever [14:36:27] no idea [14:36:29] haha [14:36:41] but i also fixed a bunch of conflicts in the test case... [14:36:52] *sigh* [14:36:58] * addshore doesnt want to have to re review the whole thing [14:36:59] but [14:36:59] meh [14:37:58] hm, there are two FIXMEs left [14:38:06] that'S nto good [14:39:43] FIXME: investigate whether we need the same kind of caching here [14:39:43] // that getKnownCurrentRevision uses (if $revId == page_latest?) [14:40:30] yea. turned that into an XXX. still open. not an immediate problem [14:40:31] / FIXME: survey callers that need to be migrated to using getSlotsQueryInfo! [14:40:36] the others are resolved. [14:41:54] oh no... two test classes where still marked as skipped. and one of them is failing :) [14:42:01] will work on that now. [14:42:18] * addshore goes back to the department retro [14:43:16] DanielK_WMDE: it hides it from your dashboard I think if it says reviewed [14:48:20] paladox: so, how many states are there, and are theys mutually exclusive? It seems there is WIP vs READY (by the owner), and REVIEWD vs UNREVIEWED (per reviewer). is that correct? [14:48:26] is it possible to t5ransfer ownership?... [14:50:24] DanielK_WMDE: not possible to transfer owner ship but you can move branches! [14:50:40] that doesn't help [14:51:08] so, to take over someone else's patch, I'd have to check them out, remove the ChangeId, and re-submit them. that loses history. [14:51:11] *sigh* [14:53:54] DanielK_WMDE: do you see a start review button on https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/421433/ ? [14:53:57] If not that’s a bug [14:54:10] (Polygerrit) [14:54:45] paladox: no, i see "mark reviewed" [14:55:03] paladox: i indeed see it in the old ui [14:55:15] Hmm [14:55:25] in the old ui i actually see both [14:55:34] Both buttons? [14:55:38] that makes no sense according to my mental model [14:55:44] yes, both buttons [14:55:59] "start review" and "makr reviewed". [14:57:17] Hmm that should be in the new ui [14:59:59] so apparently the unWIP button only exists if you use the old UI and either own the changeset or are an admin [15:00:37] that's not so bad, everyone can unwip their patches [15:00:50] still not clear how they ended up as wip in the first place [15:01:59] Tgr they were drafts! [15:02:16] if they were drafts I couldn't see them [15:02:52] anyway, a mystery for another time [15:03:27] DanielK_WMDE: https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/406595/ says TODO RELEASE NOTES, is that in another changeset? [15:03:58] tgr drafts -> wip opens them :) [15:04:13] tgr: discussed here 90 min ago: [15:33] DanielK_WMDE: it has a TODO of RELNOTES, but i guess we can followup with that? [15:05:10] tgr: i can't unwip my patches. [15:05:13] and i realyl want to. [15:05:20] Tgr it works for me [15:05:21] err, that was for paladox [15:05:21] On [15:05:30] https://gerrit.wikimedia.org/r/c/operations/puppet/+/351540/ [15:05:46] DanielK_WMDE: a admin of gerrit should be able [15:05:54] So pretty much all of ops [15:05:57] Or some of releng [15:06:42] paladox: i don't understand. i can mark my changes as WIP, but I can't undo that? Why? How is that useful? [15:07:01] why would i ever do that? [15:07:06] You should be able to in Mark them [15:07:19] hm? [15:07:26] addshore: https://gerrit.wikimedia.org/r/c/mediawiki/core/+/416465 should be good now. +2? [15:07:35] addshore: i also fixed https://gerrit.wikimedia.org/r/c/mediawiki/core/+/432053 [15:08:37] When searching for your name [15:08:41] It dosent show in gerrit [15:08:44] addshore: this one also should go in now: https://gerrit.wikimedia.org/r/c/mediawiki/core/+/425287 [15:08:55] which patches do you need an admin to un-wip? [15:09:41] legoktm: all of mine... [15:10:04] legoktm: i have no idea why they are WIP, and I don't understand how to use the feature, if I can't un-wip my own changes [15:10:57] DanielK_WMDE: were they previously drafts? [15:11:40] paladox: some where. maybe all of them were at some point. most of them definitly weren't drafts any more. [15:11:55] Ok that explains it then [15:11:57] DanielK_WMDE: example? https://gerrit.wikimedia.org/r/q/owner:daniel.kinzler%2540wikimedia.de+is:wip+status:open has a ton [15:11:59] That is why :) [15:12:08] that makes no sense to me [15:12:10] Gerrit converted all of them to wips from drafts [15:12:19] Even if they were no longer drafts! [15:12:38] DanielK_WMDE: this shows the changes now! https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/416465/28..30 [15:12:48] legoktm: yea. well, the ones that have WIP in the name can stay WIP :) [15:13:07] paladox: aaaaah, that makes more sense [15:13:11] i have made drafts before [15:13:11] paladox: great. not. [15:13:42] Yeh [15:13:47] It did it to mine [15:13:50] addshore: this is disturbing [15:14:14] DanielK_WMDE: though I can see a wip button for my [15:14:15] i guess gerrit suppresses all difs for a rebase now. [15:14:27] No it dosent [15:14:33] It’s better in polygerrit [15:14:36] paladox: i can WIP, but not un-WIP. [15:15:02] DanielK_WMDE: in polygerrit do you see a orange banner? [15:15:14] paladox: suppressing such diffs is BAD though. I can't see what I canged to accomodate the rebase. that may include substantial changes. [15:15:21] Could you take a screen shot of it please (it bring what you see on your change£ [15:15:22] ) [15:15:33] DanielK_WMDE: aha [15:15:48] This is the new rebase diff I think [15:16:21] seems broken to me :( [15:16:30] the rebase diff, i mean [15:16:49] DanielK_WMDE: one question on https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/416465/ [15:16:50] it should show diffs for all files that had a conflict resolved [15:16:51] DanielK_WMDE: see https://www.gerritcodereview.com/releases/2.15.md#New-Workflows [15:17:05] “Mark Changes Due to Rebase” [15:17:05] DanielK_WMDE: ok, everything that isn't [WIP] or [EXP] or [DNM] I set as un-wip [15:17:07] Section [15:17:30] :D [15:18:00] paladox: the behavior described there is not at all what i'm seeing. i'm not seeing anything highlighted. i'm seeing no diff at all. [15:18:09] Hmm [15:18:10] legoktm: thank you! [15:18:14] DanielK_WMDE: which change? [15:18:23] legoktm: but can you tell my why I can't un-WIP? [15:18:28] legoktm: is there any reason for that? [15:18:31] Also take a screen shot of what you see re not seeing unwip [15:19:05] DanielK_WMDE: oh honestly I have no idea. I'm just trying to be helpful [15:19:19] (using my Gerrit admin powers) [15:19:51] DanielK_WMDE: I see this https://i.imgur.com/ZQRW4zA.jpg [15:20:10] legoktm: yea, thank you :) [15:20:26] addshore: what question? [15:20:58] DanielK_WMDE: see the comment on https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/416465/ :) [15:21:52] DanielK_WMDE: do you see what’s in the screen shot on your change? :) [15:21:58] addshore: replied. the answer is "yes" :) [15:21:58] It is not marked as unwip [15:22:09] It’s text is start review [15:22:13] Which is unwip :) [15:22:27] DanielK_WMDE: can you +1 the change too? just so it doesnt look like I'm merging my own thing? :p [15:22:49] paladox: no. [15:23:09] addshore: done [15:23:32] Hmm [15:23:44] DanielK_WMDE: jenkins said no, needs a rebase [15:23:46] *lol*.... [15:24:34] paladox: https://imgur.com/a/Ardfkcf [15:24:42] paladox: maybe because of the V-1? [15:24:59] addshore: what teh fuck? [15:25:00] DanielK_WMDE: I see a "start review" button :p [15:25:19] big blue button at the top! [15:25:31] addshore: yes, but that's not the same as "ready for review", is it? [15:25:35] yup [15:25:37] I think [15:25:40] o_O [15:25:42] "start review" is the same as "set ready for review" [15:25:45] now i'm MORE confused [15:25:53] xD [15:26:23] * DanielK_WMDE is getting a headache [15:26:24] did you see the button this whole time and just think it was something else? [15:27:17] legoktm: i had seen it, but not the whole time. i was looking in the menu first. but yes, i though it was something else. tehre is also "makr reviewed". that sounds liek it's the inverse of "start review". [15:27:31] I'm not sure what mark reviewed does [15:27:34] it'S totally unclear to me how these relate to each other... [15:29:25] addshore: oh, did you merge the newNullRevision patch? [15:29:40] maybe :p [15:29:50] this is why patch chains are better ;) [15:32:07] addshore: how is beta looking? [15:32:12] alive [15:32:18] :P [15:32:27] addshore: when is the branch getting cut? [15:32:40] my day today ends in about 90 minutes... [15:32:46] DanielK_WMDE: we can do it when this patch lands [15:32:50] thcipriani is ready [15:32:59] it will be either wmf.999 or wmf.666 ;) [15:33:07] hehe [15:33:38] yea, well, we can make the branch. the question is - when will it go to testwiki? [15:33:49] how long do we want to have it on beta before that? [15:34:01] [17:32:40] my day today ends in about 90 minutes... <-- Daniel Kinzler Standard Time seems to end its days early :p [15:34:41] also, it might be easier to rebase all the patches into a stack, and then upload them with cr+2 set so you don't need to keep rebasing [15:34:47] once you have done manual testing on beta I can cut the wmf.999 branch. I wouldn't want to cut it and then pull in a bunch of backports if there are minor fixes needed. [15:34:50] legoktm: indeed [15:35:18] thcipriani: DanielK_WMDE I was going to go and test this in the garden once this patch is merged and my meeting is finished [15:35:30] \o/ [15:35:35] DanielK_WMDE: you might not be here for the branch cut ;) [15:35:38] we have the mcr checkin in an hour, too [15:35:44] DanielK_WMDE: didn't that move? [15:35:49] I think it is on monday now? no? [15:35:56] but we'll mostly be excited about getting this in, i don't expect much more from that meeting today :) [15:36:06] after that the rollout to group0 would ideally be an hour or so removed from the train, apart from that I have no preference. [15:36:22] thcipriani: yup, either an hour or so before or after [15:36:26] addshore: oh! heh, i did not notice! When did that happen? [15:36:35] DanielK_WMDE: yesterday I think [15:36:37] addshore: +1 [15:36:39] o_O [15:36:59] ok. so we are waiting for jenkins on the last patch now, right? [15:37:03] DanielK_WMDE: yup [15:37:09] and me [15:37:13] * DanielK_WMDE wibbles [15:38:20] wtf is a wibble? :P [15:39:25] Apparently different things depending on whether you're using US or British slang, at least according to https://en.wiktionary.org/wiki/wibble [15:39:32] legoktm it removes changes from your dashboard [15:39:41] so it de clutters it until a new patchset [15:39:43] ahhh [15:40:01] there's a mute button [15:40:04] and a ignore button [15:40:10] but i confuse what each other do [15:40:20] addshore: http://ars.userfriendly.org/cartoons/?id=20160905 [15:41:02] "Cartoon for Sep 05, 2016" <--- that's a lie [15:41:13] DanielK_WMDE Tgr addshore https://gerrit-review.googlesource.com/c/gerrit/+/183910 [15:41:21] DanielK_WMDE: https://www.wikidata.org/wiki/Lexeme:L2967 [15:41:25] https://gerrit-review.googlesource.com/q/owner:david.ostrovsky%2540gmail.com [15:41:31] addshore: http://ars.userfriendly.org/cartoons/?id=20010704 2001 is more like it [15:41:32] he's the one that's been doing changes to wip! [15:42:13] https://gerrit-review.googlesource.com/c/gerrit/+/184030 [15:42:41] addshore: https://www.wikidata.org/w/index.php?title=Lexeme%3AL2967&type=revision&diff=695746468&oldid=695745552 [15:42:59] well, it is also a noun, but sure [15:43:10] i feel like it might get a bit confusing before we have senses... [15:43:36] indeed [15:43:51] but... it's also a noun? really? [15:44:14] i mean, you can noun all verbs, and verb all nouns, but it weirds language... [15:44:18] English is just weird. https://www.gocomics.com/calvinandhobbes/1993/01/25 [15:44:56] anomie: exactly :D [15:45:22] DanielK_WMDE: instead of wibbling ? mind writing that email? ;) [15:46:05] addshore: to wikitech and... i forget. *sigh* [15:46:12] ops [15:47:20] Another fun quote about English: https://en.wikipedia.org/wiki/James_Nicoll#%22The_Purity_of_the_English_Language%22 [15:50:33] English is weird, but can be understood through tough thorough thought, though [15:51:19] Buffalo buffalo Buffalo buffalo buffalo buffalo Buffalo buffalo [15:51:29] addshore: what kind of commit message is https://gerrit.wikimedia.org/r/c/mediawiki/core/+/416465 >.< [15:51:58] legoktm: one that is easier to read than bd808's above sentence [15:52:44] except I already know the Buffalo buffalo joke, so it's pretty clear to me! [15:53:07] well, I already know "MCR RevisionStore, multi content mode", pretty clear to me ;) [15:53:26] touche [15:53:29] :D [15:53:32] another great word [15:56:34] addshore: do we keep T174024 open until MIGRATION_NEW is the default? [15:56:35] T174024: Implement MCR revision retrieval interface - https://phabricator.wikimedia.org/T174024 [15:56:43] or shall we close it now? [15:56:52] no i dont think so [15:56:56] IMO it is implemented [15:57:09] * DanielK_WMDE closes more tickets [16:04:24] addshore: mail sent! [16:05:12] CindyCicaleseWMF: i just marked some big blockers as resolved. the pert chart will update automatically now, right? [16:06:22] right im gonna tap out for a bit [16:06:30] already a 10 hours day today [16:07:35] oh wow. do that, then [16:12:02] addshore: The day is 17 hours old [16:12:05] You're slacking [16:17:09] Reedy: 17 hours young, maybe this day will end after 75 hours...? [16:26:11] legoktm if you approve these changes then I think they can be merged https://github.com/wikimedia/mediawiki-docker/pull/49 [16:29:03] DanielK_WMDE: so does one create a separate lexeme for verb vs. noun homographs? [16:30:32] Nemo_bis: yes. [16:30:52] ok [16:31:21] DanielK_WMDE: Great!!! Yes, once an hour. [16:31:26] tgr: this one could also go in now: https://gerrit.wikimedia.org/r/c/mediawiki/core/+/429421 [16:31:54] DanielK_WMDE: there are some unsolved problems with that [16:32:28] the one I rememeber off the top of my head is that it leaks the identity of the last editor of the page [16:33:15] tgr: i wasn't aware of any problems... I see no commens on phab or gerrit... [16:34:55] tgr: oh, you mean from https://gerrit.wikimedia.org/r/c/mediawiki/core/+/429421 ? [16:35:18] https://gerrit.wikimedia.org/r/c/mediawiki/core/+/429421/6..14/includes/page/WikiPage.php#b2805 [16:36:05] I wish we had gerrit 2.16 already, that unresolved comments view is a killer featre [16:36:23] tgr: but that issue is not new, is it? [16:36:42] the old code used the public name [16:37:00] so you couldn't really roll back if you did not see the name [16:37:12] i can't wrap my head around this right now. i'll just give it a -1 and come back to it later [16:37:19] or maybe you had to supply '' and then it rolled back all hidden users, not sure which [16:37:27] tgr 2.16 has not been released yet :) [16:37:31] oh, and I never sent my replies! [16:37:33] fun :) [16:37:36] also upstream are undecided on it being 3.0 or 2.16 [16:37:47] you compare with the raw user name, the old code compared with the user name as seen by the rollbacker [16:38:19] you fixed that for the messages but not for the comparison itself [16:38:39] I should have probably said so on the patch, sorry [16:38:50] it's hard to track what has been resolved and what hasn't [16:39:26] true [16:39:40] so, if the name is suppressed, the comparison would always fail, right? [16:40:35] I'm not sure, I think the old code might have compared with '' (which is how you see a suppressed name) and supplied that to the call and maybe that worked [16:40:55] I'll test which and leave a comment on the patch [16:41:13] i think it would have triggered an incorrect 'alreadyrolled' message. [16:42:54] I want an "attic" state. postpone for later. [16:43:08] i have a bunch of patches that i want to keep around, but am not working on right now [16:43:13] they clutter the WIP view [16:44:24] you could use hashtags I suppose [16:57:27] tgr: DanielK_WMDE well, i have done a variety of testing on beta and didn't come accross anything [16:57:36] what out of your testing list would you be most concerend about? [16:57:51] addshore: the stuff we couldn't test. VE, Flow, Echo. [16:58:04] ack [16:58:06] i can do them [16:58:16] addshore: also maybe undo, rollback, undeletion [16:58:25] excellent, thank you! [17:04:26] error code apierror-visualeditor-docserver-http O_o [17:07:53] i'll just assume visualkeditor for large pages on beta just doesnt work... [17:08:20] hm... [17:11:22] or rather, the new wikitext editor, one request i got that and another request didnt return, but editing smaller pages works fine, pretty sure it will just be some odd thing about beta being slow [17:13:01] ah, nice, captchas work. in the undo view. [17:13:16] addshore: What does the network tab tell you about the content fetch request? [17:13:37] "apierror-visualeditor-docserver-http" means "We asked MediaWiki for the content and it didn't respond.". [17:14:18] James_F: I can give you the pages I tried? :) [17:14:24] addshore: Sure. [17:14:26] DanielK_WMDE: hmmm INSERT INTO `blobs1` (blob_text) VALUES ('X') ? [17:14:56] James_F: I tried a simple 4 char edit on https://test.wikimedia.beta.wmflabs.org/w/index.php?title=Vorlage:Taxoauswahl&action=edit and also on https://deployment.wikimedia.beta.wmflabs.org/wiki/Fourier_transform?veaction=editsource using the new wikitext editor [17:15:15] the second one the ui said "Error saving data to server: Empty server response." [17:15:15] addshore: blobs1?? that doesn't look like one of mine O_O [17:15:33] addshore: where does that come from? [17:15:50] I saw it in a DBPerformance stack trace [17:16:00] but confused as to where that query is coming from? [17:16:14] addshore: oh, that's externalstore [17:16:19] and the table name is correct, i think [17:16:57] addshore: ExternalStoreDB [17:16:58] aaah yes, ExternalStoreDB, so thats not some random query coming out of nowhere :) [17:17:28] addshore: Hmm, yeah, the API request timed out after 30 seconds. [17:17:30] dud the query fail? [17:17:54] https://www.irccloud.com/pastebin/Vgsxk1mR/ [17:17:56] DanielK_WMDE: ^^ [17:18:05] just looking in logspam about what has been hapening while testing [17:18:38] "Sub-optimal transaction" doesn't sound too horrible. [17:19:10] 8.491852 delay whatever that means, between a page update and a user update *looks at code* [17:20:52] again, I'm leaning towards that being slow beta and not something that will appear in prod [17:21:43] James_F: know of any flow pages on beta i can test? :D [17:24:42] addshore: Nope. [17:25:09] *goes to check what wikis have flow pages as default talk pages* [17:25:49] is that even a thing? :P [17:28:55] addshore: MW.org [17:30:18] https://ar.wikipedia.beta.wmflabs.org/wiki/%D9%86%D9%82%D8%A7%D8%B4_%D8%A7%D9%84%D9%85%D8%B3%D8%AA%D8%AE%D8%AF%D9%85:Addshore :D [17:42:42] DanielK_WMDE: I think I'm happy with the testing and the stuff i just did, gonna make the branch [21:17:49] Krinkle: could you file a phab ticket for that failure? [21:24:27] addshore: done [22:02:09] Thanks [23:21:31] DanielK_WMDE: Do you remember what you were referring to with copyWiring() ? [23:21:32] https://github.com/wikimedia/mediawiki/blame/9bc186e021a8e386778afa3715e4f1c014b4dee6/includes/services/ServiceContainer.php#L361 [23:21:47] It seems no such method exists, not in master and not at the time the comment was merged [23:52:20] I think it might refer to an older implementation of the salvage stuff