doesn't have a class, not even one with a gibberish name [16:06:36] Although the does, and there's nothing in the

except the

[16:06:48] * RoanKattouw shakes head [16:07:05] <^demon> We can't criticize...I'll point you to the TOC on article pages :) [16:07:12] <^demon> Table with a single cell to put a div in. [16:07:19] True :) [16:22:56] RoanKattouw: do you know who I need to ping about tawiki wanting AFTv5? https://bugzilla.wikimedia.org/show_bug.cgi?id=38774 [16:23:03] for suitability etc [16:23:14] Fabrice I think [16:24:19] <^demon> RoanKattouw: Shawn already reviewed my change, it'll be in 2.5 as a result :) [16:30:06] Yay [16:36:50] <^demon> RoanKattouw: Reason I did that was that it was pointed out that 7pt is way too damn small for the search field text. [16:37:02] <^demon> Increasing the size is easier on the eyes, and might encourage people to use it more :) [16:37:08] Right [16:37:36] So re https://gerrit.wikimedia.org/r/16841 I figure we may have to style a different element given that they redid the commit msg box [16:38:13] <^demon> Yeah. Would the table's class be sufficient? Or do we really want to slap a class on the div? [16:41:10] <^demon> A more sane class name on the table would be nice. [16:41:17] <^demon> And would let you manipulate the whole thing. [16:45:20] <^demon> RoanKattouw: Ugh, the class on that table applies to some other elements as well, it's the overly generic .changeTable stuff. We really do need a class on that surrounding div. [17:13:44] RoanKattouw: Feels good to get something merged, thanks :) [17:32:13] Hi matthiasmullie chrismcmahon : Look forward to another AFT5 deployment with you guys today. I believe that RoanKattouw and Asher have approved the fix to the DB issue that caused the outage on Sunday. kaldari and bsitu are doing final code review now, so we can merge soon and push to testing. [17:33:53] hi fabriceflorin matthiasmullie and hashar have AFTv5 working on the labs beta cluster enwiki now also, I am checking it out as we speak, looks good so far [17:36:23] Thanks, chrismcmahon: Glad to hear about the good progress on labs beta cluster. Can you send me the link, please? matthiasmullie : Should I be testing the latest changes on the current prototype for now? It's probably the most practical solution for today's deployment, since all our user accounts are set up on that site. [17:38:30] fabriceflorin: http://en.wikipedia.beta.wmflabs.org/wiki/Special:ArticleFeedbackv5 I am a Steward on the labs wiki so I can manipulate user rights for you if you want [17:38:56] fabriceflorin: I'd still go for prototype, yes - labs may very well be working just fine, but I haven't been able to look into it much already, plus the resultset on prototype is larger already (we haven't yet added a lot of feedback) [17:44:14] matthiasmullie fabriceflorin agreed, but I am pleased to have a test env (in general, not just for AFTv5) that emulates prod so much better than prototype, this is pretty great :) [17:44:15] matthiasmullie: Sounds good, I will test on prototype today. Still, I am impressed by the good work you have done with hashar on the labs beta cluster! [17:44:20] chrismcmahon : could you please make Fabrice Florin (WMF) a confirmed editor on that beta site? http://en.wikipedia.beta.wmflabs.org/wiki/Special:Contributions/Fabrice_Florin_(WMF) [17:44:29] will do [17:47:41] chrismcmahon : Thanks! Also, could you please make Fabrice Florin an administrator, roll backer or reviewer on that beta site, so I can check monitoring tools? (note that this account has no WMF after its name) http://en.wikipedia.beta.wmflabs.org/wiki/User:Fabrice_Florin [17:50:04] chrismcmahon : At some point, we should create generic accounts as well for E2_Editor and E2_Monitor, so others could use test these rights as well. We should also plan to add the special aft5 user rights we now have on proto and en-wiki, which give us all oversighter rights for testing purposes. But all this can wait until later this week or next week, after we have deployed today's release. Congrats again for your good progress on this, [17:50:04] much appreciated! [17:50:24] Is there a reason https://gerrit.wikimedia.org/r/#/c/16967/ isn't approved? [17:50:25] fabriceflorin: I've made you a 'confirmed user' and 'rollbacker' on beta enwiki, if there are other privs you want, let me know http://en.wikipedia.beta.wmflabs.org/wiki/Special:ListGroupRights [17:53:09] Hi chrismcmahon : Thanks! So Fabrice Florin (WMF) is just a confirmed editor, but Fabrice Florin is both an editor and a rollabacker, right? Thanks again for setting this up, so I can test the view for each user group separately. [17:55:45] fabriceflorin: ha! actually, I missed that you have 2 accounts. right now 'Fabrice Florin (WMF)' is 'confirmed user' and 'rollbacker' and 'Fabrice Florin' has no group rights yet. [17:56:33] Hi matthiasmullie and chrismcmahon : I have pasted the list of items to be deployed today on this ether pad, so we can track it on proto, testing and production. http://etherpad.wikimedia.org/tFyLAurf0Z Feel free to add any additional items we might have overlooked, as well as any important test notes that we might want to track throughout deployment. [18:00:37] Hi Matthias: When testing as an editor on prototype, I get CTA3 - Take a survey, which was supposed to be disabled across all user groups, since the survey is over. This is not a deal-breaker, but if it's easy to remove it, I would do this now. [18:03:24] matthiasmullie: I am consistently getting CTA3 as an editor on prototype, 4 times out of 4. What these editors should be seeing is CTA5, which invites them to check the feedback page (instead of taking an old survey). I think this is sufficiently annoying that we should take the time to remove this issue. [18:04:02] fabriceflorin: the "can this CTA be displayed" checks along with the fallback routine ("if not CTA1, try CTA2, if not, try CTA3, …") doesn't seem to take the bucketing into account (which only serves as a starting point to start the CTA-routine) [18:08:09] matthiasmullie: So do you think that this will be fixed when we get to testing and production? Or do we need to make a fix now? I can reproduce this issue as a monitor and oversighter as well. Going forward, I'm thinking we may want to simplify the bucketing, if it proves to be fragile, to only show CTA5 to all editors and only CTA1 to all readers (except if it's a protected page, when would show CTA2). This would remove CTA4 from [18:08:10] circulation for now, but might fix the problem temporarily, if it occurs on production as well. [18:13:39] fabrice: could you try again on prototype (after clearing you browser's cache) [18:15:13] it's a quick fix, making sure CTA3 is never hit - will refactor the code by next release to also incorporate another check with bucketing [18:20:25] matthiasmullie: The CTA3 is gone now, but I now get CTA4 (sign up or login), even though I'm a registered user and monitor on prototype. This is not a deal-breaker for this deployment, because very few admins use the feedback tool for posting comments, but it should be fixed by next week. [18:20:55] matthiasmullie: I am trying again as a reader and editor next. Will clear my cache each time. [18:26:36] matthiasmullie: Good news and bad news. As a logged out reader, I get CTA1 every time. However, I am still getting the Survey CTA 3 as a confirmed editor on prototype -- with the Fabrice Florin (WMF) account. (I may actually be auto-confirmed, if this makes a difference, though I'm not sure anymore). http://prototype.wikimedia.org/release-en/User:Fabrice_Florin_(WMF) [18:28:16] matthiasmullie: Given that we are running out of time, let's go with this solution for now, because the readers are the most important user group. But let's try to fix (and test thoroughly) these CTA issues this week. To reduce complexity, I am prepared to lose the bucketing altogether, and just make the CTA dependent on user rights, with only one option per user group. Will update the Bugzilla ticket now. [18:34:28] matthiasmullie: When I test the CTAs on prototype in Chrome incognito mode (no cookies or preferences), I get CTA5 (links to feedback page). This is not a show-stopper for deployment, but one more indication that the bucketing seems broken on prototype. I will add all these comments on the Bugzilla ticket, for the record. [18:36:25] fabriceflorin: what's broken with seeing cta5? [18:36:48] on another browser/incognito, you probably got bucketed to cta5 :) [18:39:15] matthiasmullie: Good point. I take back my comment about CTAs being 'broken' on prototype, as you are correct that we did specify that CTA5 would be in rotation with CTA1 for readers. Sorry for being inaccurate. Let's discuss this tomorrow, I will send you an invite for a 1+1, so we can figure out a simple solution for all these CTA issues. [18:39:56] sounds great [18:46:33] so, how would some people feel about doing some code review on Echo? [18:46:58] It's my 20% day today [18:47:00] I can do some [18:47:06] I'm about to come into the office [18:47:24] I stayed home this morning because I cut my finger while making breakfast :S [18:48:41] werdna : So glad you're pushing the envelope on Echo. Once the dust settles with AFT and Page Curation, I plan to start getting more involved in notifications … In the meantime, can't wait to see what you guys come up with your new prototype! [18:48:48] RoanKattouw: cut... but you still have it attached to your hand, right? [18:49:22] Yeah it didn't cut it /off/ :) [18:49:25] Nothing serious [18:49:33] Just some blood on the countertop and stuff [18:49:48] I mean I cut myself with a cheese knife, those aren't very sharp [18:50:24] cuts with dull knives are the worst [18:51:01] Well it's not a dull knife either, it's fairly new. But it's not a Stanley knife either [19:08:58] fabriceflorin: chrismcmahon: the changes have been pushed to test [19:09:13] Sounds good. I'm on it! [19:09:46] check [19:16:37] matthiasmullie: fabriceflorin when I try to submit feedback for Sparrow on testwiki I am getting the message "Form submission error" [19:17:09] matthiasmullie: The talk page link on testing is showing up several lines below the title, for both readers and editors, with a strange link in between: "Talk:Golden-crowned Sparrow [19:17:10] http://test.wikipedia.org/s/rw5 [19:17:10] View reader feedback » " [19:17:30] I can reproduce this on Chrome, Firefox and Safari: http://test.wikipedia.org/wiki/Talk:Golden-crowned_Sparrow [19:21:20] not getting "Form submission error" any more, seems that was a fluke of some sort [19:21:40] fabriceflorin: that http://test.wikipedia.org/s/rw5 is added by another extension (ShortUrl), which is also the reason the aft link is pushed down [19:22:44] that extension isn't active on enwiki, only test [19:23:08] Cool. Just checking. I am not seeing any issues that I can test directly on testing. All the oversighter fixes can only be tested on production. Have we verified that the change to My Contribs data table is working as intended on testing? or can this only be tested on production right now? [19:23:31] (and some other exotic wikis: hiwiki, orwiki, tawiki, tawikibooks, tawikinews, tawikiquote, tawikisource, tawiktionary, testwiki, test2wiki) [19:24:22] fabriceflorin: asher has already created the index, so we're good to re-enable aft on enwiki [19:24:41] chrismcmahon: I couldn't reproduce that either, will keep an eye out for it though [19:24:44] matthiasmullie : BTW, the CTAs all performed as intended during my test on testing. In all 3 user modes I checked, no CTA3 was displayed. [19:24:56] ok, sounds good [19:25:20] I've been in IE7 and things look OK as far as I can see [19:25:45] chrismcmahon : Cool. So unless you have issues on testing, I think we are ready to move to production. [19:26:12] no issues found [19:27:04] okay - pushing to production, will take a while :) [19:27:13] * chrismcmahon has just about learned that any issue found, particularly with IE7, within the first couple of minutes post-deploy, is probably server caching glitch of some sort [19:27:37] Thanks, matthiasmullie - will go grab a bite now, will be back at my desk in 10-15 mins. [19:30:00] ;D [19:44:07] RoanKattouw: are you in and lunched yet? [19:47:10] In but not luncched [19:47:23] okay, poke me when you're lunched [19:59:52] Lunching at 1:30 [20:00:10] werdna: If you wanna talk to me before that go ahead [20:10:24] fabriceflorin: chrismcmahon: this week's changes have been deployed to production [20:10:38] thanks matthiasmullie [20:10:44] RoanKattouw: can I come down? [20:11:00] Sure [20:11:03] Sounds great. Starting test now. [20:11:11] I'm waiting for Rob to be done with his intervie [20:14:29] * Danny_B|backup thought the office is only one floor but large [20:24:33] matthiasmullie : The talk page link doesn't appear for me in Chrome Incognito, as a logged out user (with no cookies). http://en.wikipedia.org/wiki/Talk:Golden-crowned_Sparrow [20:25:14] matthiasmullie fabriceflorin I'm seeing some page errors in IE7 I did not see on testwiki, and some unusual behavior on the central page for Sparrow. narrowing that down if I can [20:28:07] chrismcmahon: It may be possible that we are not fully deployed. I am not seeing the talk page link at all on Chrome (regular or Incognito) or Firefox, but I am seeing it on Safari. [20:30:46] interestingly, IE7 has a page error upon login at the line 'if (wgUserGroups.indexOf("sysop")...', I don't think that is AFTv5 code [20:31:02] (nice that crossbrowsertesting calls the debugger for you) [20:31:08] moving on to AFT pages... [20:32:31] fabriceflorin: that's odd, I'm seeing it just fine, don't really have an idea what could be causing that [20:33:06] chrismcmahon: that doesn't look like aft code indeed, botting into IE7 VM as well [20:34:52] yeah, I'm getting page errors from a 'speedy deletion' bit of js as well [20:35:21] something about 'wa_window', happens only occasionally [20:36:30] matthiasmullie : This could be due to the fact that not all servers have caught on to the new deployment. It's often the case that changes are not propagated fully for hours after deployment, in my experience. In some cases, Roan used to also clear the server caches for features involving cookies. For now, I can still reproduce the absence of a talk page link on Firefox and Safari, but only for readers. Not a show-stopper for this week's [20:36:31] deployment. [20:36:50] I'm going to test the oversighter issues now. [20:40:34] fabriceflorin: I'm gonna blame caching indeed, as I'm consistently seeing the link [20:41:23] chrismcmahon: what page are you seeing those weird errors? I'm not seeing anything on my VM [20:42:17] matthiasmullie: every page. [20:50:11] matthiasmullie: hiding/unhiding in IE7 seems flaky. the grey overlay doesn't always appear (but unlike fluffernutter's issue, the unhide/oversight buttons still work) [20:53:15] I can confirm that the oversight request email that was sent via the Central Feedback page was properly sent to the OTRS queue. In fact, [20:54:03] In fact, Fluffernutter oversighted that item minutes after I posted it, which is good news for two reasons: 1) she most likely got the email; 2) she is able to oversight [20:54:56] I can also confirm that the links in the email message and in the suppression log are now correctly going to the feedback page (with the oversight-related link on top), not to the permalink page. [20:55:52] ok, I think I'm going to call IE7 flaky but close enough [20:55:57] So I think we can declare these two oversight issues 'mission accomplished' - thanks, Matthias! We still have the larger oversighter javascript issue, but at least Fluffernutter can do her job, which is the most important thing. [20:58:21] chrismcmahon: ok, I'm able to reproduce the hidden overlay-issue on IE7 - thanks [21:00:04] matthiasmullie: that's close enough to fluffernutter's issue that it might be a clue for later [21:01:00] matthiasmullie: I just quit and restarted Firefox and Chrome and can now see the talk page links as a reader on all my browsers. So that issue is taken care of, must have been a slow server update. [21:08:25] I'm still getting some strange behavior on the Higgs Boson page, probably due to a slow server - It doesn't show a talk page link or feedback form on Chrome, but it shows up on Firefox and Safari: http://en.wikipedia.org/wiki/Higgs_boson Probably a server issue, not an application issue, but I am reporting it just in case. [21:09:45] checking.... [21:10:34] fabriceflorin: I have a working 'view reader feedback' link off the Higgs Boson page in Chrome [21:10:52] yes, me too [21:10:56] talk page I mean [21:11:12] Thanks, Chris. Sounds like we're good on that front, then. Slow server issue, not our problem. [21:11:51] Overall, we seem to be fine with today's deployment from my standpoint. chrismcmahon, do you have any critical issues on your end? [21:12:26] fabriceflorin: no issues except non-fatal flakiness in IE7 [21:13:09] that is, the issues I saw for IE7 are all fixed by reloading the page or or clicking a link to a different view, that sort of thing. nothing broken. [21:15:00] matthiasmullie chrismcmahon : OK, let's call it a wrap, then! Thanks for all your great work today, as always. Next AFT deployment will be Mon. Aug. 13, using the platform train. Hopefully that will just be a casual check, with no drama and less work than these manual deployments. To be safe, I will as tewwy to hold Tuesday morning windows for us on Aug. 14 and Aug. 28, in case something terrible happens that needs to be fixed manually the [21:15:00] next day. [21:16:29] sounds good, see you next time! :) [21:16:54] sounds good! [21:19:35] Yes, of course I'll hold the Tuesday morning window open. [21:23:45] tewwy: Splendid, thank you! [21:24:14] matthiasmullie: Nice work. Get some rest, and let's catch up Wed. at 10:15am, right after our daily SCRUM. [21:24:50] sounds good, talk to you tomorrow! [21:34:12] RoanKattouw: can you look at https://gerrit.wikimedia.org/r/#/c/17125/ [21:35:30] preilly: To review the extension or just to approve the submodule change? [21:35:46] RoanKattouw: approve the submodule change [21:35:57] Done [21:36:47] RoanKattouw / other deployers -- if you're available, could you look at https://gerrit.wikimedia.org/r/#/c/17111/ + syncdir the extension directory to testwiki? [21:37:04] RoanKattouw: and New patchset: preilly; "Add GeoData" [mediawiki/tools/release] (master) - https://gerrit.wikimedia.org/r/17126 [21:38:13] ori-l: Approved [21:38:20] RoanKattouw: hero :) [21:38:23] preilly: Will approve that one. You're about to update on fenari, right? [21:38:37] RoanKattouw: yes [21:38:52] OK then could you take care of Ori's second request (update E3Experiments on fenari)? [21:39:00] I just approved a submodule change for ti [21:43:18] ^^ preilly [21:43:32] Tip of the day: Don't put "This is the official repository of " in your github repo description. It doesn't really work well from a forked repo. [21:43:48] ori-l: okay [21:43:54] preilly: thanks [22:00:18] Getting "Error: invalid magic word 'coordinates'" all over testwiki right now [22:00:47] ! [22:00:48] ok, you just typed an exclamation mark with no meaning in the channel, good job. If you want to see a list of all keys, check !botbrain [22:00:55] * RoanKattouw slaps wm-bot [22:03:21] The TimedMediaHandler deployment is running long [22:07:21] StevenW: AaronSchulz is working on the problem now [22:09:30] I love the list of ffmpeg codecs in TMH readme [22:09:51] I can already imagine people uploading YUV4MPEG files to Wikimedia servers... [22:10:24] robla: Thx^3! [22:12:22] hmm, why does this "Error: invalid magic word 'coordinates'" happen even after localisation cache rebuild? [22:14:19] StevenW: test.wikipedia.org is fixed now. by the way...are you doing testing there, or are you reporting something you heard from someone else? [22:14:33] I was visiting to do testing [22:14:55] we need to rename test to avoid confusion. it's kind of a crappy place to do testing for anything other than deployments [22:15:15] test2.wikipedia.org is more appropriate [22:15:19] Oh yeah, this isn't normal pre-deployment testing for UX requirements etc. [22:15:26] We do that elsewhere [22:16:00] if you're looking to test "does this work on the production cluster", test2 is where its at [22:16:56] robla: huh, i didn't know about test2. my bad. [22:23:56] j^: what do you mean by simplified? [22:52:57] Catchable fatal error: Argument 1 passed to ArticleFeedbackv5Render::grayMask() must be an array, null given, called in /usr/local/apache/common-local/php-1.20wmf8/extensions/ArticleFeedbackv5/Artic [22:52:58] leFeedbackv5.render.php on line 162 and defined in /usr/local/apache/common-local/php-1.20wmf8/extensions/ArticleFeedbackv5/ArticleFeedbackv5.render.php on line 310 [22:53:04] ^ can we get these fixed, please! [22:53:46] who wants to go to Bourbon & Branch? [22:53:50] tonight [22:54:42] nice catch Reedy I think that might be the source of some flaky behavior we have not been able to pin down [22:55:02] I think there's a bug logged for it.. [22:55:55] https://bugzilla.wikimedia.org/show_bug.cgi?id=38359 [22:55:56] !38499 [22:55:58] maybe [22:56:06] Also, there's https://bugzilla.wikimedia.org/show_bug.cgi?id=38440 [22:56:13] Reedy, could you repeat that for matthiasmullie_ ? [22:56:20] (who just joined) [22:56:46] matthiasmullie_: Catchable fatal error: Argument 1 passed to ArticleFeedbackv5Render::grayMask() must be an array, null given, called in /usr/local/apache/common-local/php-1.20wmf8/extensions/ArticleFeedbackv5/Artic [22:56:46] leFeedbackv5.render.php on line 162 and defined in /usr/local/apache/common-local/php-1.20wmf8/extensions/ArticleFeedbackv5/ArticleFeedbackv5.render.php on line 310 [22:56:53] ^ just noting that hasn't been fixed yet [22:57:22] which is bug 38499, and there's also 38440, but I'm not sure if that's still occurring [22:58:02] 38440 is a similar condition, but afaik has not affected behavior. but we have seen intermittent strange behavior with the grayMask [23:03:19] There's no one CC'd on these AFTv5 bugs... [23:03:24] I just assigned them both to him [23:07:51] . [23:08:00] . [23:11:03] ． [23:11:19] • [23:12:45] ･ [23:15:19] Reedy: in between my connection issues, I saw you the bugs you mentionned ;)