[08:11:37] h [08:11:38] ell [08:11:39] o [08:11:44] yo [10:01:21] ok [10:04:20] Hi, I am a gsoc aspirant and have compiled my proposal for one of the projects. https://www.mediawiki.org/wiki/User:Karthikprasad/gsoc2012proposal [10:04:25] i would truly appreciate if you could kindly go through and suggest some corrections/additions. [10:04:28] thanks [14:32:57] New patchset: Hashar; "checkstyle target to run PHP CodeSniffer analyses" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/4272 [14:33:47] New review: Hashar; "(no comment)" [integration/jenkins] (master); V: 1 C: 2; - https://gerrit.wikimedia.org/r/4272 [14:33:50] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/4272 [14:45:02] AND THERE. https://www.mediawiki.org/wiki/Review_queue is far more up-to-date now. [14:45:05] Bam. [14:46:53] hexmode: ^ in case that helps you out [14:47:05] <^demon> :) [14:47:13] :) [15:15:13] New patchset: Hashar; "checkstyle job requires dirs to be created in ws" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/4274 [15:15:47] New review: Hashar; "(no comment)" [integration/jenkins] (master); V: 1 C: 2; - https://gerrit.wikimedia.org/r/4274 [15:15:49] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/4274 [15:24:53] New patchset: Hashar; "PHP CodeSniffer job. not in main workflow yet" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/4275 [15:25:14] New review: Hashar; "(no comment)" [integration/jenkins] (master); V: 1 C: 2; - https://gerrit.wikimedia.org/r/4275 [15:25:17] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/4275 [15:26:17] morning RoanKattouw [15:26:47] Morning [15:31:30] Krinkle: So what are you working on today? [15:31:51] right now I'm handling incoming mail and mw.org RL2 talk page [15:32:02] checking task management in a few minutes [15:32:23] KO [15:32:26] Pretty much the same here [15:32:34] I was looking at Inbox (44) this morning [15:33:08] <^demon> Good news. Gerrit 2.3 should be released today [15:33:12] <^demon> *fingers crossed* [15:36:59] RoanKattouw: gerrit-dev on labs was upgraded, right? [15:37:12] * Krinkle tries to find out why the layout looks uglier than it used to [15:37:15] Yes, ^demon did that [15:37:26] ok, so the css is no longer compatible, as expected [15:37:28] <^demon> gerrit-dev is on 2.3rc0. [15:37:36] <^demon> If 2.3 goes final today, I plan to upgrade again. [15:37:45] <^demon> And start testing 2.3 more thoroughly for a production upgrade. [15:47:09] RoanKattouw: k, I'm thru with the primary mail, taking a quick foodbreak and then map out todays tasks? [15:47:20] Sure [16:10:43] rsterbin: Do you need any of the changes that were submitted into Gerrit yesterday to be included in today's deployment? I haven't looked at them yet because I was on a plane yesterday, just noticed them now [16:20:50] RoanKattouw: 'm back [16:21:09] k [16:22:46] https://www.mediawiki.org/wiki/RL2 [16:22:55] https://www.mediawiki.org/wiki/ResourceLoader/Version_2_Design_Specification/Task_management [16:24:05] OK so I should probably finish the labs setup and fix that JS-messes-with-prefs-display bug [16:24:38] k [16:24:47] I'll get on error handling in the gadget manager [16:25:04] Then maybe some of the other backend items like adding the skin and position properties [16:31:45] OMG, I have to get through Wikitech (53) first [16:31:50] Forgot about that :D [16:40:05] Oh touché, I've to (137) at that [16:40:31] I didn't take the mailing lists yet, I'll read them in the weekends :P [16:40:57] although some of the headlines I'm scanning now seem attractive [16:44:40] New patchset: Hashar; "rethink git checkout and fingerprinting" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/4284 [16:45:34] New review: Hashar; "merging directly, this is already in production." [integration/jenkins] (master); V: 1 C: 2; - https://gerrit.wikimedia.org/r/4284 [16:45:36] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/4284 [17:00:46] RoanKattouw: do you have a sec to talk about the db changes with the AFT release today? [17:02:54] Sure [17:02:57] rsterbin: Back now [17:03:13] i've got the sql at the bottom of that alter.sql file [17:03:27] Hello RoanKattouw, yoni_omniti, rsterbin, werdna and everyone! Are we ready for today's deployment of the abuse filter on AFT5? [17:04:01] but i've been thinking that it might be smarter not to change the column name, but rather to add the column, copy the values, and then delete the old one later [17:04:22] fabriceflorin: yep [17:04:24] It doesn't really matter what we do, if I have to add a column I'm already altering the table [17:04:37] Hmm but I think you're right [17:04:44] It's probably better to do it this way [17:05:01] Cool. For today's first test of the abuse filter on AFT5, we will only have one filter, which rejects any comments that are ALL CAPS. Our primary goal for today's release is to test that the Abuse filter and the Article Feedback extensions are well integrated and to measure their joint impact on performance. [17:05:03] since the code starts writing to the new column, you'd have to update the db and the code at the same time [17:05:11] Yeah exactly [17:05:33] do you want updated sql, or do you just want to do it? [17:05:47] I can probably just do it [17:05:50] Lemme check the table size again [17:05:50] cool [17:05:58] In future releases in coming weeks, we will have more filters and more actions, but still be limited to 0.5% of the English Encyclopedia. We plan to go for a wide release of Article Feedback in late May/early June. In the meantime, you can learn more about this abuse filter on our AFT5 feature requirements page: [17:05:58] http://www.mediawiki.org/wiki/Article_feedback/Version_5/Feature_Requirements#Abuse.2FSpam_Filters [17:09:18] mysql> alter table aft_article_feedback add column af_form_id INTEGER UNSIGNED NOT NULL DEFAULT 0, add column af_experiment varchar(32) NULL, add index af_experiment (af_experiment); [17:09:21] Is what I just ran on testwiki [17:09:33] cool [17:10:08] UPDATE aft_article_feedback SET af_form_id = af_bucket_id; ? [17:10:25] after that the other updates should be good to run [17:10:27] Ah yes that too [17:10:37] i figured; just checking :) [17:10:56] Good catch, I forgot [17:11:01] and ran the other UPDATEs first [17:11:03] RoanKattouw: There's a bug I'm randomly hitting in gmail, maybe you've encountered it too. I'm filtering mails (e.g. "label:wikitech-l is:unread"), starting at the back and navigating up from there (e.g. from 100, to 99, to 98 etc.). But then sometimes at random I'm suddenly at 30 or 20, realizing I somehow skipped a whole bunch [17:11:05] Ran yours then reran the others [17:11:05] sounds familiar? [17:11:11] cool [17:11:17] Krinkle: Yes, can happen if you reply or someone else does [17:11:23] bumps it up in the timestamp order [17:11:44] I mean like skipping 30 at once [17:11:54] rsterbin: So I need to deploy everything that's in master now, nothing more nothing less, right? [17:11:58] hmm [17:12:14] did elizabeth's bug fixes from yesterday make it in? [17:12:26] one sec, i can check myself, duh [17:12:33] I was on a plane yesterday so probably not [17:13:24] <^demon> Krinkle: I've had intermittent unread bugs in gmail for about a week now. [17:13:39] yeah, those haven't been reviewed yet [17:13:57] everything submitted to AFT on gerrit is intended for today's release [17:14:03] OK [17:14:06] I'll review those real quick then [17:14:08] it could be that a new mail came in that triggered the bug, but definitely not as many mails that it "skipped" [17:14:10] thx [17:16:28] hi robla hexmode and me [17:16:37] howdy [17:16:38] Thanks, rsterbin and RoanKattouw. Are the two fixes from Elizabeth about the abuse filter or the feedback page? [17:17:10] hey, Nikerabbit I already pinged you about that log rewrite bug... [17:17:14] hi 20% people [17:17:24] fabriceflorin: those are bug fixes for the feedback page [17:17:45] oh, right, and the new icons [17:17:58] do you want RoanKattouw to not push them? [17:18:42] it's Wednesday, so, RoanKattouw and raindrift (Ian)? [17:19:06] sumanah: I'm Tuesdays now, I need to update that on the wiki page, sorry [17:19:17] ah, thanks Roan. [17:20:02] hexmode: there are dozens of them [17:20:52] Nikerabbit: I was talking about the one where I was looking for you today [17:21:02] * hexmode checks scrollback [17:21:43] Nikerabbit: https://bugzilla.wikimedia.org/35689 [17:23:33] Hi rsterbin, I think it would be great if we pushed the new icons and feedback page fixes, as long as they have been tested on your end as well. They work on my end. [17:23:51] does Ian generally not come into IRC by around now? [17:24:06] fabriceflorin: elizabeth and i have tested locally and on prototype [17:24:07] hexmode: could I ask you to ping Ian later today to ask him about his 20%? [17:24:20] sumanah: np [17:24:53] RoanKattouw: fixed: http://www.mediawiki.org/wiki/20_percent [17:25:18] hexmode: if it is easy it is easy [17:25:22] thx robla [17:25:34] Splendid. Just wanted to dot the i's, so we are clear that these features are also being deployed, once [17:25:47] robla: you probably saw my wikitech-l email re extension review? there's a teensy bit in there for you to respond to [17:26:48] …. once RoanKattouw completes his code review. Also note that once we have deployed these new abuse filter features and AFT5 bug fixes, we will want to deploy the AFT4 changes which DarTar requested, ideally in today's deployment. [17:27:32] RoanKattouw: https://gerrit.wikimedia.org/r/#change,4286 [17:29:11] rsterbin: https://gerrit.wikimedia.org/r/#change,4186 needs fixin [17:30:11] RoanKattouw: Elizabeth is fixing it [17:30:17] OK [17:30:29] if I can figure out how to get the fix into gerrit [17:30:55] rsterbin: Could you help her with that? ---^^ [17:31:16] attempting to — i've never done it before, either ;) [17:32:08] re [17:32:32] Thanks, RoanKattouw, rsterbin and auroraeosrose. Since we have a limited deployment window, is it realistic to do this fix now? Or should we push the abuse filter features now, so we don't miss our deadline? (Also note that we need Roan to make some changes about AFT4 in the same deployment window). [17:32:59] can we please prioritize the data fixes? [17:33:11] these are the only time critical changes [17:33:18] I think I'm just gonna deploy the things that were already reviewed [17:33:33] RoanKattouw: agree [17:33:38] To sum up: our top two priorities are to push the abuse filter changes and to do make the AFT4 changes requested by Dario last week. [17:33:41] Submitting changes while your reviewer is 32,000 ft above the North Atlantic and then expecting to have them in production the next day is .... not realistic :) [17:33:47] feedback page bug fixes aren't critical [17:33:55] Yes, RoanKattouw, I agree as well. [17:34:05] rsterbin: ditto [17:34:16] OK, generating a patch from git then [17:35:27] the most important issue with AFT5 is the change in the aft_article_feedback table, the AFT4 changes have lower priority [17:37:26] Thanks, DarTar. Can you point us to the specific Bugzilla tickets for the AFT5 changes you need, so we can make sure we have them in this release? Also, please share the link to what needs to be done on AFT4, once we've deployed AFT5. [17:38:17] https://bugzilla.wikimedia.org/show_bug.cgi?id=35619 [17:38:38] that's the only critical AFT5 fix [17:39:46] that fix is in the patch Roan is working on now [17:39:56] for AFT4, we don't have a ticket but RoanKattouw is aware of them: (1) disabling entirely the cookie-based throttling for AFT4 CTAs (2) appending additional data (article_title|rev_id) to all AFT4 events in the log [17:40:48] Thanks, rsterbin, glad we were able to include that revision in this release: Also: https://gerrit.wikimedia.org/r/4030 and https://gerrit.wikimedia.org/r/4031 [17:41:34] right [17:42:03] bugzilla:35619 also requires updating the aft_article_feedback tables in production, but that's something we can probably do later [17:42:21] Thanks, DarTar. After we deploy AFT5, let's make sure to remind RoanKattouw about these specific AFT4 requirements, as he has a lot on his plate today, and may be slightly jet-lagged ;o) [17:42:25] rsterbin: you said you already had the queries handy, right? [17:42:48] Yeah thanks for reminding me [17:43:08] I am sitting right behind him and will pester him until I get them pushed ;) [17:43:10] DarTar: Roan ran them already; we're prepped and ready to go [17:43:18] sweet [17:46:28] Cool, nice work you guys! I guess that's one benefit of RoanKattouw having just flown back from Europe: it gets him to start his day much earlier than the rest of us ;o) [17:53:27] fabriceflorin, rsterbin: OK the code is now on testwiki, please test there [17:53:37] I'll do the schema change on enwiki in the meantime [17:53:37] word [17:53:45] awesome [17:53:57] RoanKattouw: I have a note to self: "in the future, all gadgets will need to be RL2-compatible -- start publicizing this -- will start sometime after the 1.19 deployment (check w Roan for when)" -- this is at some point after Gadgets 2.0 comes out, right? so, it's maybe sometime in August? [17:54:12] sumanah: This starts when Gadgets 2.0 is deployed [17:54:36] I think we should announce this widely once we have that software in a public testing ground and have invited public testing and stuff [17:54:49] fabriceflorin, yoni_omniti: don't forget to get your "before" submit times on enwiki to test AF+AFT5 performance [17:55:25] guillom: ^ heads-up on that. But according to the roadmap we probably won't be able to make that announcement till May at the earliest [17:56:04] "** In April, set up a first working setup in WMF labs [Roan, Timo]" [17:56:41] We sort of started to announce it with the 1.19 deployment [17:56:46] True [17:56:58] But I'm available to help with a wider communications campaign when the time comes [17:57:04] Rock. [17:58:07] preilly: Could you paste that link to that Gerrit overview in PM? [18:00:45] Thanks, RoanKattouw, testing now. Rsterbin, thanks for the reminder to clock our performance times before and after on production. Will do that next. [18:01:21] RoanKattouw: I'm getting 500 errors on test [18:02:10] 220 PHP Warning: require_once(/home/wikipedia/common/multiversion/MWVersion.php) [function.require-once]: failed to open stream: No such file or direct [18:02:12] ory in /usr/local/apache/common-local/live-1.5/MWVersion.php on line 12 [18:02:17] Is that what you see? [18:02:34] I am also getting this error message when I test AFT5 abuse filter on testing: "Form submission error." The URL is: http://test.wikipedia.org/wiki/Golden-crowned_Sparrow [18:03:10] This error message was triggered when I entered text in ALL-CAPS in the URL above. [18:03:47] RoanKattow: In firebug, I get the html for a page with the title "Wikimedia Error" [18:04:11] fabriceflorin: what about without all-caps? [18:05:31] same error w/o all caps [18:05:33] db? [18:05:36] rsterbin, I am getting the same error message whether or not I type ALL-CAPS. I'm on Safari, haven't tested on other browsers yet. Does anyone need a screenshot, or can you reproduce on your end? [18:05:44] Yeah things are screwed up [18:05:50] yeah, i can reproduce it, too [18:05:56] Like seriously screwed up [18:06:32] is there a db complaint in the error log? [18:06:44] No, that's not it [18:07:22] It appears to be sending testwiki requests to machines that aren't set up to handle them [18:07:27] That's a very weird operational issue [18:07:40] Are you guys still getting these 500s now? For which URLs? [18:08:06] yes [18:08:14] yes [18:08:14] it's the aft submit api request [18:08:31] http://test.wikipedia.org/w/api.php [18:08:34] Oh, hmm, wait a second [18:08:49] I might have forgotten to add the .flagging.php file [18:09:02] Yup, I did [18:09:05] head, meet desk [18:09:11] hehe [18:09:38] rsterbin: regarding the schema change, I now see in aft_article_feedback a new af_form_id field and an af_experiment VARCHAR field, is the plan to drop af_bucket_id once the data in the tables has been updated? [18:09:49] Yes [18:09:51] DarTar: that's right [18:09:57] ok [18:11:57] I'll have to leave around 11.45 (hosting speaker for seminar at 12), if you guys need me it has to be now or after 1.30 PDT [18:12:11] OK, are these 500s fixed now? [18:12:14] I added the missing file [18:12:24] yes [18:12:29] Yy [18:12:31] Yay [18:13:56] RoanKattouw: can you add that config var so abuse filter gets turned on in test? [18:14:07] I entered text in all-caps on testing, but did not get any message telling me that my feedback was rejected. Is that supposed to happen? Have we not yet added the 'all-caps' shouting filter on testing? http://test.wikipedia.org/wiki/Golden-crowned_Sparrow [18:14:13] rsterbin: Varname? [18:14:25] $wgArticleFeedbackv5AbuseFiltering [18:15:41] d'oh. i don't have permission to edit abuse filters on test [18:16:06] fabriceflorin: i need to get permission to add it on test [18:16:17] amusingly, i already have permission on enwiki [18:16:25] Hi DarTar, can you tell me what needs to be done to confirm that the AFT5 change you requested is working when we deploy it to production? Also, are the AFT4 changes done earlier by Roan working for you? [18:16:38] I haven't done the AFT4 changes yet [18:17:59] Hi rsterbin, I don't know how to give you permission to edit the abuse filter on testing. I believe that Werdna gave us these permissions on prototype and en-wiki, but didn't think of giving them to us on testing. Roan, do you have the ability to do that, or should we deploy to production instead? [18:18:46] RoanKattouw, thanks for clarifying that the AFT4 changes have not yet been done. I must have misunderstood that information ;o) [18:19:07] I'll give rsterbin permissions [18:19:08] rsterbin: Username? [18:19:11] rsterbin [18:19:14] rsterbin: (BTW I enabled that var) [18:19:18] thx [18:20:12] rsterbin: Done [18:20:20] thx [18:20:25] fabriceflorin: we need to make sure that the bucket data is captured in the new af_experiment field [18:20:48] where bucket data equals what we track in the AFT5 log [18:21:34] also af_form_id should now get the same values as af_bucket_id [18:22:11] Thanks, DarTar. How can I test this on my end? Is there a link to a tool server that can let me confirm that for you? Or would I need access to a special DB? [18:22:49] I don't have access to the DB on test, I can only test this once the changes are in production [18:23:00] RoanKattouw: I'm getting this message on prevention — Your comment violates the http://test.wikipedia.org/wiki/<articleFeedbackv5-error-abuse-link>. Please revise it. [18:23:14] it's an old version; could you clear the message cache? [18:23:21] Oh, right [18:23:26] Sorry abuot that [18:23:31] no worries :) [18:24:21] Rebuilding, this'll take a little while [18:26:25] rsterbin: do you have access to the test DB? [18:26:44] rsterbin: OK should work now, it's gone past en [18:26:46] no, i don't [18:27:07] all the analysis and a team of 20 volunteers coders are on hold until we have this data [18:27:32] RoanKattow: no dice just yet; could it be a js caching issue? [18:27:38] hmm [18:29:16] RoanKattouw, how about going straight to production with these fixes, rather than wasting time trying to configure testing, which could take for ever? If the release doesn't work, we can always revert. [18:30:14] rsterbin: Try now? [18:30:20] ok... [18:30:21] let me make another suggestion, it seems we're getting stuck with abuse filter and feedback page fixes that are totally not time-critical, if they don't work they can easily wait one week, the data fixes cannot (unless we revise the entire metrics plan) [18:30:24] is there something special I need to do to mentor someone for GSOC? [18:30:59] RoanKattouw: still the wrong message [18:31:36] hmph [18:31:40] <^demon> Ryan_Lane: Spare time :) [18:31:42] RoanKattouw, this time the abuse filter caught my all-caps message, but it is displaying the wrong message: "Your comment violates the <articleFeedbackv5-error-abuse-linktext>. Please revise it." [18:31:49] do I need to sign up somewhere? [18:31:56] DarTar: the feedback page fixes are already out of this release; the abuse filter would just need to be turned off in prod [18:32:38] rsterbin: There's a typo in the message key, articleFeedbackv5-error-abuse-linktext doesn't exist [18:32:49] Perhaps you meant articlefeedbackv5-error-abuse-linktext [18:32:55] Wait [18:33:01] then… why does it work properly on prototype? [18:33:03] just saying that i'll be gone in 10 and nobody knows if we're collecting data that is usable or not [18:33:08] wtf [18:33:11] Ryan_Lane: sign up on th gsoc page [18:33:16] there's a section for mentors [18:33:31] rsterbin: I have no idea, but there's a capitalization difference here I can tell you [18:33:32] RoanKattouw: just for sanity, can you check the i18n file? [18:33:38] articlefeedbackv5-error-abuse-linktext vs <articleFeedbackv5-error-abuse-linktext> [18:33:40] I checked [18:33:47] The i18n file has a lowercase f [18:33:50] Hi RoanKattouw, can you extend this deployment today to make sure that DarTar's AFT5 request goes through? [18:33:55] Yes of course [18:35:00] apergos: which gsoc page? :( [18:35:09] sec [18:35:40] http://www.mediawiki.org/wiki/Summer_of_Code_2012 [18:35:54] I put this link in the ops etherpad a while back [18:36:02] I know, no one reads what I write :-P :-P [18:36:13] you're on that already [18:36:14] no. I mean how do I do so on google's site? [18:36:17] ah [18:36:41] I assume they are usin the same system as the last coupl years [18:36:45] lemme dig aroun [18:36:46] d [18:36:52] rsterbin: There was a capitalization difference in your code, I've fixed it live, does it work now? [18:36:56] where's sumanah when you need her? [18:37:04] have we been accepted as an org yet? [18:37:13] I think so [18:37:17] She's in a meeting I think [18:37:20] Thanks, RoanKattouw, we really appreciate your willingness to extend the development as needed to include DarTar's requests, which he has been patiently waiting for weeks ;o) [18:38:28] there's also a mailing list which is mostly a waste of time [18:39:39] http://www.google-melange.com/gsoc/homepage/google/gsoc2012 - that's the gsoc site [18:39:47] http://www.google-melange.com/gsoc/homepage/google/gsoc2012 [18:39:48] melange [18:39:50] RoanKattouw and rsterbin, I am still getting the same old error message on testing, even after the live fix. [18:40:02] i no loner see errors on test, but I'm able to submit a feedback comment in all caps [18:40:03] yeah there's a link below the student register that says register as a mentor [18:40:06] *longer [18:40:10] I hope melange is better than it used to be, [18:40:31] not really ;) [18:40:35] bummer [18:40:50] do we have the filter configured in test? [18:40:55] I don't think we do [18:41:28] rsterbin: the fix to entirely disable option2 and option3 is already in production, or isn't it? [18:41:34] Thanks, guys, I just got the right message this time: "Your post has been rejected by a software filter that suggests it may have violated Wikipedia's feedback guidelines. Please revise your post and try again." But the link to the feedback guidelines is wrong: it should be "http://en.wikipedia.org/wiki/Wikipedia:Feedback_guidelines", not "http://test.wikipedia.org/wiki/<articleFeedbackv5-error-abuse-link>", [18:41:56] fabriceflorin: i've just found the problem on the message [18:42:00] committing now... [18:42:02] Oh rawr that one too [18:42:04] I fixed the other one live [18:42:15] But apparently both message keys are affected [18:43:26] Thanks, RoanKattouw and Rsterbin. Let me know when you think you have fixed the link. Once we confirm that it works, I recommend we move to production, to test this further -- and deploy DarTar's AFT5 change. [18:43:27] RoanKattouw: https://gerrit.wikimedia.org/r/4290 fixes both [18:43:44] Thnks [18:44:46] np [18:48:28] OK, this works for me, both the right message and the right link are now displayed when I shout. So I recommend we move to production next, if that works for you. [18:49:45] fabriceflorin: agreed [18:50:26] ok folks, gotta go - will be back around 1.30-ish [18:51:38] OK, let's go to prod then [18:52:57] Running the deployment now, I'll tell you when it's finished running [18:53:00] lol, for a minute there I thought Dario was referring to MediaWIki release numers [18:53:01] This'll take like 10 mins [18:53:09] "ok folks, gotta go - will be back around 1.30-ish" [18:53:20] lol [18:53:45] Thanks, RoanKattow! Please let us know when we are live on production, so rsterbin can add the abuse filter for shouting. Also, DarTar will check the AFT5 changes on production at 1:30pm PT, as well as the AFT4 changes, on the assumption that we will have deployed them all by then. Sound good? [18:54:13] sounds good to me [18:54:23] OK [18:57:50] Time to do our clocking experiments on production for the "before" condition, folks. [18:59:13] i did mine earlier [18:59:31] firebug helpfully provides the execution time on api calls [19:00:37] I got times in the 250ms to 500ms range [19:01:42] I just tested the new code and it works as intended on Chrome: en.wikipedia.org/wiki/Golden-crowned_Sparrow [19:02:15] rsterbin: same here [19:03:09] fabriceflorin: what do you mean by "new code"? i haven't added the filter yet. [19:04:03] also, i don't think Roan's said prod is done. [19:04:42] It's still running [19:05:19] It may be a caching issue on Chrome. When I tried to post all caps at this URL, I got the message: en.wikipedia.org/wiki/Golden-crowned_Sparrow - But now I cannot reproduce after emptying my cache. [19:05:33] <^demon> RoanKattouw: I'm going to rename "VisualEditor Team" to "extension-VisualEditor" for consistency with my new arbitrary naming conventions. [19:06:01] OK [19:06:46] In any case, the lag time I am getting is about 200 milliseconds for the BEFORE condition on production, using the WMF servers. I will try now with Verizon Mifi to see if it changes it. [19:07:14] OK it's done syncing [19:07:26] purge proxies ? [19:10:08] RoanKattouw: I got something for you :-D [19:10:24] yoni_omniti: On it [19:10:32] RoanKattouw: Introducing MeetBot, in use on #jenkins . Let you takes meeting notes in IRC channel and generate an automatic minutes report [19:10:39] yoni_omniti: Done [19:10:44] RoanKattouw: http://wiki.debian.org/MeetBot / example output : http://meetings.jenkins-ci.org/jenkins/2012/jenkins.2012-04-04-18.01.html [19:10:51] hashar: OK, cool, but why me? :) [19:10:54] RoanKattouw: thanks! [19:11:18] RoanKattouw: config var is set? [19:11:33] RoanKattouw: cause you are in SF, a member of the feature team, and the awesome dude interested in crazy stuff :] [19:11:52] (sorry about the ArticleFeedback interruption) [19:12:02] rsterbin: On testwiki yes [19:12:02] the filter needs to be created [19:12:03] Not on enwiki yet [19:12:05] Want me to set it there too? [19:12:14] yep [19:12:17] OK [19:12:32] yoni_omniti: i just created the filter [19:12:39] thanks [19:12:46] let me know when to test [19:13:02] OK enabled [19:13:04] RoanKattouw: i've got a change I need to push out to MobileFrontend - will it mess you up if i do a sync-file on fenari? [19:13:10] yoni_omniti: good to test [19:13:21] awjr: No go ahead [19:13:24] ty [19:13:59] appears to be working correctly for me, tested several variations [19:14:11] the links and messages appear to be correct too [19:14:34] did fabrice d/c? [19:14:56] hmmmm [19:15:12] fabrice is experiencing technical difficulties ;) [19:15:42] welcome back [19:15:43] i'm getting the correct messages, too [19:15:50] ok looks good to me [19:15:52] times are now 1-1.5s [19:16:03] i am getting same response times, below 800ms [19:16:08] ouch [19:16:55] Thanks, yoni_ominiti. Sorry for disappearing, but my Verizon MiFi test disabled my access to the IRC server. For the BEFORE timing measurements, I was generally getting between 100 ms and 200 ms on all browsers, whether or not I use WMF's ethernet and Verizon MiFi. So I will continue my test with WMF server instead. [19:17:41] fabriceflorin: you'll want to test on the same network as you tested before deployment [19:18:19] we're testing server execution time, not network request/response [19:19:21] * RoanKattouw points to the comments in the HTML of each response [19:19:42] RoanKattouw: its an ajax request, so this doesn't help us [19:19:45] unfortunately [19:19:50] Oh, right :S [19:21:07] Hi guys, I am able to get the same response times on production AFTER deployment (between 100-200 ms), but am getting the wrong messages from the server: "Warning: An automated filter has identified this edit as removing a large amount of content. While some material on Wikipedia should be removed, such as vandalism and false or libelous information, we discourage removal of content without a valid reason. If you believe that your removal is [19:21:08] constructive, please go to the bottom of this page and press 'Save page' to continue (be sure to enter an edit summary if you have not done so already). If you did not remove a large amount of content, and you believe that this edit is constructive, please report this error. [19:21:09] Your action has triggered the Abuse Filter An automated filter has identified this edit as potentially unconstructive, and it has been disallowed. If this edit is constructive, please report this error. [19:21:09] " [19:21:25] thats not what I'm getting [19:21:29] I'm seeing the correct message [19:21:30] me either [19:21:35] clear browser cace [19:21:44] cache* [19:21:46] This message appears for me on both Chrome and Firefox. I will clear my caches again now. [19:21:48] <---- can't type [19:21:48] what comment text are you trying? [19:24:01] OK, after clearing cache on Firefox, I got the correct message and response was good as well (150 milliseconds). [19:24:54] hmmm [19:25:02] I'm running rsterbin type numbers here [19:25:06] 1.05s [19:25:43] actually more like 800ms -> 1.05s but that range [19:27:05] yeah, i've gotten a couple between 800ms and 1s, too [19:29:37] Works for me on Chrome as well. I get the same performance after deployment as before.But I was incorrect earlier when I was stating that the wait time was 150 milliseconds - I misread my stopwatch, it really was 1.5 seconds all along. But that was both before and after, so I am not seeing a serious degradation on my end. However, this is not really a scientific test. [19:30:41] fyi, fabriceflorin & yoni_omniti: here's the log of hits on the new filter — http://en.wikipedia.org/w/index.php?title=Special:AbuseLog&wpSearchFilter=458 [19:33:51] fabriceflorin, i don't think that weird message you got was a fluke. i just spotted this in the logs: http://en.wikipedia.org/wiki/Special:AbuseLog/6633058 [19:33:55]