[13:22:01] (CR) Cmcmahon: [C: 2] "documentation" [qa/browsertests] - https://gerrit.wikimedia.org/r/95591 (owner: Zfilipin) [14:16:15] (PS1) Zfilipin: Added links to Jenkins jobs [selenium] - https://gerrit.wikimedia.org/r/96023 [14:18:45] hashar: how do I find https://gerrit.wikimedia.org/r/#/c/96023/ at https://git.wikimedia.org/tree/mediawiki%2Fselenium ? [14:21:08] if I edit the url directly then it works [14:21:09] https://git.wikimedia.org/blob/mediawiki%2Fselenium/233eabe179a75079b399cded6f458d1c4d8d917a/README.md [14:21:22] zeljkof: append the reference at the end of the URL, escaping slashes with %2F [14:21:23] but is there a way to get to that page from the website [14:21:24] ie: http://git.wikimedia.org/log/mediawiki%2Fselenium/changes%2F23%2F96023%2F1 [14:21:37] to always point to `master` http://git.wikimedia.org/log/mediawiki%2Fselenium/master [14:21:50] and three versions before master: http://git.wikimedia.org/log/mediawiki%2Fselenium/master^^^ [14:22:03] tag: v [14:22:03] http://git.wikimedia.org/log/mediawiki%2Fselenium/0.1.7 [14:22:08] yw [14:22:11] (CR) Zfilipin: "How the new page looks like: https://git.wikimedia.org/blob/mediawiki%2Fselenium/233eabe179a75079b399cded6f458d1c4d8d917a/README.md" [selenium] - https://gerrit.wikimedia.org/r/96023 (owner: Zfilipin) [14:22:18] https://git.wikimedia.org/blob/mediawiki%2Fselenium/master/README.md [14:22:56] hashar: thanks, so in short, without editing the URL manually, there is not way to get there*? [14:23:23] and maybe you can render README.md from change 96023 with https://git.wikimedia.org/blob/mediawiki%2Fselenium/changes%2F23%2F96023%2F1/README.md [14:23:46] I don't think there is a straightforward way [14:23:59] hashar: thanks [14:23:59] gitblit resolves everything to commit sha1 :/ [14:24:15] editing URL is simple enough, I just thought there will be a way from the web interface [14:50:57] (CR) Zfilipin: "Relevant bug: https://bugzilla.wikimedia.org/show_bug.cgi?id=57188" [selenium] - https://gerrit.wikimedia.org/r/92650 (owner: Zfilipin) [14:58:57] (CR) Zfilipin: "Relevant bug: https://bugzilla.wikimedia.org/show_bug.cgi?id=57188" [selenium] - https://gerrit.wikimedia.org/r/93443 (owner: Zfilipin) [15:30:46] (PS1) Mwjames: Remove Scribunto from smw-core unit test suite [integration/jenkins-job-builder-config] - https://gerrit.wikimedia.org/r/96032 [16:16:44] (CR) Jeroen De Dauw: [C: 1] Remove Scribunto from smw-core unit test suite [integration/jenkins-job-builder-config] - https://gerrit.wikimedia.org/r/96032 (owner: Mwjames) [16:38:53] (PS1) Addshore: Switch Query and QueryEngine to checks only [integration/zuul-config] - https://gerrit.wikimedia.org/r/96043 [16:39:36] (PS1) Addshore: Remove dependancies that are not on gerrit [integration/jenkins-job-builder-config] - https://gerrit.wikimedia.org/r/96044 [16:40:29] (PS2) Aude: Remove dependancies that are not on gerrit [integration/jenkins-job-builder-config] - https://gerrit.wikimedia.org/r/96044 (owner: Addshore) [16:40:51] (CR) KartikMistry: [C: 1] "LGTM." [selenium] - https://gerrit.wikimedia.org/r/96023 (owner: Zfilipin) [16:41:46] (CR) Aude: [C: 1] Remove dependencies that are not on gerrit [integration/jenkins-job-builder-config] - https://gerrit.wikimedia.org/r/96044 (owner: Addshore) [16:42:08] (CR) Addshore: "As an alternative please see I0be4f86d2b862ee83ea1d6fa8a812d83f2b6cce8 and I866568379ed67d22343a86d565b249b7f763f01f" [integration/zuul-config] - https://gerrit.wikimedia.org/r/93428 (owner: Jeroen De Dauw) [16:42:15] (CR) Addshore: "As an alternative please see I0be4f86d2b862ee83ea1d6fa8a812d83f2b6cce8 and I866568379ed67d22343a86d565b249b7f763f01f" [integration/jenkins-job-builder-config] - https://gerrit.wikimedia.org/r/95616 (owner: Jeroen De Dauw) [16:42:18] (CR) Aude: [C: 1] Switch Query and QueryEngine to checks only [integration/zuul-config] - https://gerrit.wikimedia.org/r/96043 (owner: Addshore) [16:44:33] hashar ^^ do you think that might be possible? [16:44:38] (PS1) Zfilipin: MobileFrontend Jenkins jobs run just features/scenarios with appropriate tag [selenium] - https://gerrit.wikimedia.org/r/96048 [16:45:02] (PS2) Zfilipin: MobileFrontend Jenkins jobs run just features/scenarios with appropriate tag [selenium] - https://gerrit.wikimedia.org/r/96048 [16:46:02] addshore: no clue, I am off. Will follow up either tonight or tomorrow [16:46:15] okay :) [16:47:28] (CR) Cmcmahon: [C: 2] "maintenance" [selenium] - https://gerrit.wikimedia.org/r/96048 (owner: Zfilipin) [16:50:30] (CR) Jhall: [C: 1] Added links to Jenkins jobs [selenium] - https://gerrit.wikimedia.org/r/96023 (owner: Zfilipin) [17:23:08] (PS1) Zfilipin: All MobileFrontend Jenkins jobs should send e-mail notification to the same addresses [selenium] - https://gerrit.wikimedia.org/r/96058 [18:50:49] hi JeffHall did that -1 on https://gerrit.wikimedia.org/r/#/c/96062/ make sense? [18:51:10] Yes, it did - I'll work on that today! [18:51:51] Rummana also had a suggestion for adding a Scenario for a user that is not logged in, so I'll plan on adding that case as well [18:51:52] JeffHall: cool, I think the existing examples should be enough to go on, but we could pair program in a bit. [18:52:06] OK, thanks Chris [18:53:06] JeffHall ryasmeen actually, a test for a not-logged-in user would not be very useful I think. for one thing, VE on beta is only available for logged-in users. for another thing, it would be duplicate effort. [18:53:31] JeffHall: the reasoning behind always checking the diff view after editing is: [18:54:30] JeffHall: the contentEditable surface in VE is called "CE" in local jargon-- but CE is not the data that is saved when you hit "Save page"... [18:55:11] JeffHall: instead, the contents of CE are loaded into memory... [18:56:24] called "data JeffHall: a big risk for VE is that the contents of CE differ from the contents of Data Model or DM (I remembered!) [18:57:20] JeffHall: and we can check CE vs. DM in the diff view [18:57:40] chrismcmahon: Ah, OK - that is interesting. I'll take a look at how that works and expand the new test to incorporate that Diff view part of the workflow [18:58:23] JeffHall: we have seen cases in the past where CE is correct, but DM is wrong. It is DM data that is saved to wiki page in the final step. [19:04:10] JeffHall: I'm grabbing a quick lunch, back shortly. [20:37:48] hi JeffHall [20:38:07] Hi chrismcmahon [20:38:29] JeffHall: we should go over the workflow for updating a branch of code [20:39:04] JeffHall: my bible for that is https://www.mediawiki.org/wiki/Gerrit/Getting_started [20:39:21] chrismcmahon: Agreed - I'm think I'm messing up somewhere... [20:39:27] JeffHall: by which I mean... [20:40:02] JeffHall: you made this and it was awesome: https://gerrit.wikimedia.org/r/#/c/96062/ [20:41:20] JeffHall: but then instead of updating that branch you ended up making a new branch with a new id and a new commit msg: https://gerrit.wikimedia.org/r/#/c/96083/ [20:41:58] chrismcmahon: Yeah, I should have used that "--amend" option - thanks for reminding me about that wiki page! [20:42:14] (CR) Jeroen De Dauw: [C: 1] Switch Query and QueryEngine to checks only [integration/zuul-config] - https://gerrit.wikimedia.org/r/96043 (owner: Addshore) [20:43:07] JeffHall: exactly. unless you use --amend you'll create a new branch, which is not what you want in this case. also, you should keep the same commit message, describing the code, and make any notes about updates describing the patch sets in the comments [20:43:40] (CR) Jeroen De Dauw: "This should only be merged in case Jenkins will +1 rather then +2 after this change." [integration/zuul-config] - https://gerrit.wikimedia.org/r/96043 (owner: Addshore) [20:45:07] JeffHall: so I think we'll probably merge 96083 and abandon 96062, but let me make sure everything is sane before I actually do it. [20:46:19] chrismcmahon: OK, thanks. Sorry for the confusion - still trying to get totally comfortable with the Gerrit workflow [20:46:39] JeffHall: not at all, I did the same thing myself, more than once! [20:46:56] JeffHall: thus I recognized the symptoms :-) [20:47:16] (CR) Jeroen De Dauw: [C: 1] Remove dependencies that are not on gerrit [integration/jenkins-job-builder-config] - https://gerrit.wikimedia.org/r/96044 (owner: Addshore) [20:55:51] JeffHall: and thanks for re-using existing steps! I'm glad you used 'Then This is a new line should appear in the diff view', that is a nice feature of Cucumber [20:56:39] chrismcmahon: Yeah, I'm a big believer in DRY, so I'm all about code reuse! [21:04:35] (PS1) Cmcmahon: UW sometimes takes time to process. Discussed on QA mail list [qa/browsertests] - https://gerrit.wikimedia.org/r/96146 [21:11:13] JeffHall: so when you review more code, for example https://gerrit.wikimedia.org/r/#/c/96146/ maybe, don't set any of the fields under "Verified", just the ones under "Code-Review" [21:11:57] JeffHall: the "Verified" fields will be set by Jenkins [21:14:32] chrismcmahon: Yeah, Ċ½eljko already slapped my hand about using the "Verfied" fields, so I know how to behave myself properly around those settings :) [21:15:28] JeffHall: yep, you can see we've made these mistakes ourselves [21:39:29] Has anyone else noticed login difficulties with automated tests in the qa-browsertests repo? [21:39:40] The tests want to login from a URL with this form: [21:39:42] https://test2.wikipedia.org/wiki/Main_PageSpecial:UserLogin [21:40:03] But the correct page is really located at: [21:40:04] https://test2.wikipedia.org/wiki/Special:UserLogin [21:40:26] I just noticed this today, but haven't tracked down the source of the problem yet... [21:41:32] JeffHall: Zeljko did some changes, that probably got mangled, I was looking at that also [21:41:37] JeffHall: lemme fix that [21:41:50] Great, thanks Chris [21:42:16] JeffHall: thanks for checking, I saw the failed logins, hadn't seen the mangled URL yet [21:46:33] JeffHall: actually, I am not seeing a mangled URL in the Jenkins jobs for test2, just the wrong password. where do you see the Main_PageSpecial:UserLogin issue? [21:48:01] chrismcmahon: Ah, I think this is stupid user error - I believe I exported the wrong value for MEDIAWIKI_URL - will check that now... [21:48:33] JeffHall: sounds like it 'export MEDIAWIKI_URL=http://test2.wikipedia.org/wiki/' should do it. [21:48:53] chrismcmahon: Yeah, that fixed it! [21:53:28] hello there [21:53:39] * hashar hates timezone differences [22:00:03] hi hashar you're awake late today [22:02:44] argh, I hate flakiness. JeffHall if you want to see what happens in Chrome when the beta warning popup fails to be dismissed properly, https://wmf.ci.cloudbees.com/job/VisualEditor-en.wikipedia.beta.wmflabs.org-linux-chrome/lastCompletedBuild/testReport/(root)/VisualEditor%20Bullets,%20Numbering/check_strings_for_bullets_and_numbering__outline_example_____Numbering_______/ [22:07:55] chrismcmahon: Interesting - I noticed a similar error running a test locally with Chrome, where Firefox has no problem with the same step [22:10:09] JeffHall: Chrome will tank if it thinks that an element is "in front" of the element you want to interact with. the W3C is actually debating right now how to handle this sort of thing in WebDriver [22:11:32] (CR) Jhall: [C: 1] UW sometimes takes time to process. Discussed on QA mail list [qa/browsertests] - https://gerrit.wikimedia.org/r/96146 (owner: Cmcmahon) [22:18:43] chrismcmahon: i got my weekly meetings on monday evening [22:27:31] chrismcmahon: if you are around, I have spot an issue with ULS browser tests and phantomjs [22:27:34] https://bugzilla.wikimedia.org/show_bug.cgi?id=57101 [22:28:09] related to how browsers normalize font-familly: 'foo',bar differently :( [23:33:51] hey JeffHall I noticed you added yourself to https://bugzilla.wikimedia.org/show_bug.cgi?id=56567 . I think the issue has fixed itself over the weekend. I'm not sure the root cause, but I'm not seeing Chrome thrash like it did since roughly Friday. [23:34:37] chrismcmahon: OK, that makes sense then - I was trying to find a way to repro the problem reported in the bug, and wasn't having much success! [23:35:53] JeffHall: yeah, it is bizarre, but in a good way. I've been banging my head on that problem for a long time, and now it is no longer a problem. It could have been a change in mediawiki, or a change in Chrome31/ChromeDriver2.6, or some combination. [23:36:23] JeffHall: but it is AWFULLY nice to see all those Chrome builds finally go green. [23:37:18] chrismcmahon: Yeah, definitely good news! [23:40:55] JeffHall: I think I have 3 BZ tickets open for that Chrome issue, I'll be closing them this week.