[08:18:50] morning! [08:24:26] yo [08:28:44] hi phuedx [08:28:51] hey joakino, how're you? [08:29:10] phuedx: i'm writing a huge comment here https://phabricator.wikimedia.org/T100296 to enlighten the problem and get to any solution [08:29:22] joakino: sure [08:29:57] i had a brief chat with greg and i'm unsure if stable branches are the way forward -- on the other hand, our ci infrastructure is lacking, so anything that gives us more confidence is a bonus [08:30:28] greg's point, which is completely valid, is that master should be releasable [08:30:42] our problem is that, we're not convinced that that is true [08:38:59] phuedx: done https://phabricator.wikimedia.org/T100296 [08:39:19] phuedx: releasable master makes sense, if we have a development branch that gets deployed to staging [08:39:52] phuedx: i've tried to explain at the end that the solution is both ways, we don't care, whatever separates our development concerns with our production concerns [08:40:01] will work [08:42:04] joakino: so something that occurred to me on saturday was feature flags [08:42:21] which would require no change to our branching/releasing model [08:42:31] and would give us complete power over when things are turned on [08:42:38] at the cost of code complexity [08:42:51] ^ given that our code is already overly complex, it's a significant cost [08:42:55] phuedx: we already have a lot of that hah [08:42:56] yea [08:43:07] will add to phab in a sec [08:43:14] just catching up on /normal/ email ?/ [08:43:15] :/ [08:43:31] phuedx: that is true for features, but you can't differentiate between different states on the feature (ie old patches vs new patches) [08:43:42] ye me too [08:44:51] joakino: hexplain -- not sure that i'm following [08:45:51] phuedx: feature flags don't solve our problem of not-signed-off patches being deployed [08:49:23] they sorta do -- the browse prototype has been on beta for ~2 weeks now [08:50:36] phuedx: you mean accesible on beta labs but not on production [08:50:42] yup [08:50:51] and that's how we've been getting design review [08:51:12] phuedx: the thing is after the feature is enabled in both staging and production you can't control the patches that get merged to it with only one branch [08:51:24] true enough [08:51:34] i'm not suggesting that it's /the/ solution [08:51:44] phuedx: agreed it's the way forward [08:51:52] more than anything, i want a mechanism to put backpressure on devs [08:51:56] not on releng [08:52:05] phuedx: how are you feature-flagging CSS? just resource loader not loading the modules? [08:52:11] yeah [08:52:23] 👍 [08:52:26] which, admittedly, means ++$complexity [08:52:35] but otoh means small, focussed modules [11:07:01] father's group time [11:16:35] enjoy! [13:34:34] yo [16:08:20] phuedx: ping [16:09:05] bgerstle: pong [16:09:26] bgerstle: assume that i don't read scrollback [16:09:27] KABOOM [16:09:41] bgerstle: coreyfloyd: oh hai [16:09:51] phuedx: i want to add a param to mobileview that effectively sets "page" to the site's main page [16:09:52] hey mhurd [16:10:05] phuedx: so we don't need to ask siteinfo for it [16:10:39] so "gimme the main page for the site that i'm hitting if this is set, otherwise do a thing"? [16:10:51] yep [16:11:13] mhurd: sup [16:11:14] add an if statement at the top of the request handler that checks for the param, let's call it siteMainPage [16:11:34] phuedx: if present, the request's target is the title of the site's main page [16:11:39] otherwise, fall back to the "page" parameter [16:11:40] mhurd: pulled some things into 58 - pop off what you like… [16:11:51] if siteMainPage and page are both specified, 400 [16:11:55] (if we were using HTTP codes) [16:12:01] coreyfloyd: bgerstle: batcave in a bit to catch up? [16:12:19] mhurd: if you want, also have standup in an hour [16:12:32] mhurd: can’t this minute… but sent an email with the release details [16:12:34] bgerstle whats wrong with page: 'Main+Page', [16:13:28] mhurd: I’m basically in meetings from now until 5… trying to cram 15 minutes of work in before the next one [16:13:41] joakino, bgerstle: also see #wikimedia-dev [16:13:46] coreyfloyd: oh wow. [16:14:02] coreyfloyd: that's nuts [16:14:05] mhurd: bgerstle we should think about what we want to demo at the showcase though… [16:14:26] mhurd: bgerstle we each pick a thing? [16:14:46] coreyfloyd: sure. i guess the description editing for me? [16:14:58] coreyfloyd: there's a showcase today?? [16:15:14] if we're all demoing hacks, i say cancel it [16:15:18] mhurd: bgerstle yeah 3pm EDT… [16:15:30] mhurd: bgerstle yeah we should probably not demo the hacks… [16:15:33] joakino: do you have anything to demo? [16:15:37] probably what we shipped [16:15:39] in 130 [16:15:55] not sure that we've got a whole bunch to demo from web [16:16:12] coreyfloyd: why not? [16:16:34] coreyfloyd: everyone already saw them... [16:16:46] coreyfloyd: will think of something else... [16:17:05] phuedx: just my app for the people that didn't come to Lyon [16:20:03] mhurd: yeah - we just demoed a week ago, so thinking we should demo what we are shipping [16:20:37] mhurd: i guess we can demo if we have nothing else, but would try other things first… [16:21:09] coreyfloyd: cool [16:22:30] phuedx: random mobile web question. I hear that MobileFrontend uses cookies. do you know what for? [16:27:56] gilles: could you send an email to the list? i've got meetings for the next ~day and i might not respond as quickly as you'd like [16:28:00] someone else might :) [16:28:14] alright, thanks [16:34:59] coreyfloyd: could i hear tasha in the background? [16:35:49] phuedx|MEETINZ: yeah - i had it muted so not sure how [16:43:03] coreyfloyd: https://media3.giphy.com/media/862U5nF51e4da/200.gif [16:43:59] dbrant: https://etherpad.wikimedia.org/p/search-logging [16:52:18] joakino: i need more unicorn shirts [17:14:53] FlorianSW: apologies i owe you lots of replies [17:16:45] jdlrobson2: really? I must say, i haven't checked my e-mails very often during the weekend, i'm currently working on moving my wiki to another host and do some maintenance work, that's why i'm a bit busy :] [17:17:15] okay FlorianSW :) mostly phabricator bugs and patches but they're on my to do list :) [17:17:31] jdlrobson2: ah, great! :) [17:29:12] jdlrobson2: yt? [17:30:19] yo samuel d. there is no hangout in the calendar cc phuedx|MEETINZ [17:30:31] yup [17:30:33] on it [17:31:02] phuedx|MEETINZ: Where's this meeting? I don't see a video link? [17:31:49] interestingly i can't create a hangout :/ [17:31:58] via calendar.google.com [17:32:12] phuedx|MEETINZ: yup see pm :) [17:32:40] oh shoot we have a meeting? what room phuedx|MEETINZ [17:32:48] kaldari iiii iiii [17:33:03] I'm confused :) [17:33:16] * joakino renamed to joakino|MEETINZ_NOHANGOUT_WTF [17:33:29] phuedx|MEETINZ: that's very bad thought lidering [17:33:38] thought lidoing [17:33:39] ? [17:34:01] it's ringing on my phone dammit [17:34:06] is there a link [17:34:23] jdlrobson2: i have ipad/phone and laptop dancing [17:34:33] OMGGGZZ [17:34:41] also time to switch my calendar back to usa time.. [18:03:03] phuedx|MEETINZ: would you like to revisit https://gerrit.wikimedia.org/r/#/c/210649/ ? :) [18:07:31] * FlorianSW is now under 1000 unread e-mails *yuhu* [18:12:04] JonKatz: yt? [18:12:18] phuedx|MEETINZ supp! [18:12:32] we should sync before i go on holly hols [18:14:20] i'm free for next 30. [18:14:26] phuedx|MEETINZ ^ [18:16:03] phuedx|MEETINZ or we could schedule, as it looks like you might be in MEETINZ--I'll set something up. [18:20:13] JonKatz_: now? [18:20:19] i've got a little time [18:21:09] phuedx|MEETINZ just sent you an invite and dug into something else. let's do that unless the suggested time tomorrow is bad [18:21:35] JonKatz_: cool [18:26:22] bgerstle: formatversion=2 isn't temporary, but the results it produces are subject to change as API modules are updated to return cleaner results. Like mobileview returning a proper boolean [18:40:26] greg-g: now that i've removed the word 'labs' may i get a normal response? :p [18:40:28] https://phabricator.wikimedia.org/T100296#1325052 [18:41:01] joakino: will do when I have time to read fully and reply, sorry, mondays are busy [18:41:15] greg-g: ok thanks. no worries [18:42:08] derpyderp derp [19:19:00] jdlrobson: do you have a little time, I need some suggestions on unit-tests? [19:24:07] hey codezee sure. i can give you my time at 1.30pm my time - is that too late for you? i'm in meetings right now [19:25:35] jdlrobson: I guess that'd be an hour from now? [19:27:04] https://www.wikidata.org/wiki/Help:Description "A description describes and disambiguates a label using information that is free of controversy and bias" [19:31:03] codezee: yup does that work fro you? [19:31:18] jdlrobson: sure, np, I'll wait :) [19:43:13] joakino's related pages is implemented extensions/GettingStarted/MoreLikePageSuggester.php [19:48:35] ... and it turns out MoreLikePageSuggester.php is just making a search query for "morelike:Australia". Who knew Cirrus search had a "morelike:" keyword? [19:57:00] it's awesome :D [20:34:54] @codezee hey [20:34:58] i'm free now [20:36:43] jdlrobson: yes, a few questions regarding tests [20:37:01] codezee: sure ask away :) [20:37:16] you see at http://git.wikimedia.org/blob/mediawiki%2Fextensions%2FWikidataPageBanner.git/8f067abadcb60fdc649667fbcaefcc9637dafc24/includes%2FWikidataPageBanner.hooks.php [20:37:27] a major part of the code deals with the banner(image) [20:38:14] so my question is, while writing a test, I'll need to add a test image and see if its working or not, and there's the problem, its a test, and how am I to know if the image is present on the wiki or not at all? [20:38:54] I mean the basic question is, how to deal with tests involving media like here [20:39:39] codezee: can you point me to the test? [20:39:43] or some code you've written? [20:39:49] you can post it on gist.github.com [20:41:17] its a very rough version but here https://dpaste.de/DEpT [20:42:06] * jdlrobson looks [20:42:45] codezee: what exactly are you trying to test? [20:42:48] addDefaultBanner ? [20:43:27] jdlrobson: yes, the one's responsible for adding the default banner, I'm trying to check for that [20:43:58] umm so i suspect you want to use stubs. have you used those before? [20:44:03] it's been a while since i have [20:44:19] codezee: which patchset is addDefaultBanner defined? is the code in master? [20:44:39] jdlrobson: yes, its there, in the link I provided above [20:45:13] jdlrobson: I've heard of but not used stubs, if you could point to a working example of it, I'll take some references [20:46:17] codezee: take a look at MobileFrontend phpunit/api/ApiMobileViewTest.php [20:46:29] and how MockApiMobileView works [20:46:46] basically you extend a class and override certain functions with your own that do predictable things [20:47:40] jdlrobson: you mean we extend the class to be tested? [20:47:47] yup [20:48:47] jdlrobson: so does that mean we're basically writing a sort of mirror version of code with static values, instead of variables to test [20:49:45] so you should mock the things that are difficult to test [20:50:03] e.g. if the class has a function getImageUrl which looks for a file and turns it into a thumbnail [20:50:17] you just replace it with return "/img/mock.jpg" [20:50:36] so it avoids running that code and forcing you to have to think about it [20:51:12] jdlrobson: alright I think I can try it out now and let you know in case of an issue, thanks [20:51:34] good luck! :) [20:53:02] jdlrobson: also, there are a number of patches, but should we first concentrate on https://gerrit.wikimedia.org/r/#/c/212512/ because its before the new one's ? [20:56:30] codezee: yup looking through them all now :) [20:56:53] codezee: so that one - the syntax you use for srcset doesn't quite match the spec for the srcset attribute [20:58:01] codezee: http://www.w3.org/html/wg/drafts/html/master/semantics.html#image-candidate-string [20:59:54] jdlrobson: I see that, I'll fix the format [21:00:31] http://html5hub.com/srcset-attribute-solving-responsive-image-dilemma/ < see this for ideas [21:02:52] turns out I wasn't aware that an actual srcset attribute did exist, I was trying to implement randomly :P [21:04:31] codezee: haha :) [21:04:42] codezee: always look out for standards when you can [21:04:51] in future you may want to use srcset rather than data-srcset too [21:05:10] using srcset would mean you wouldn't need any js for certain browsers to do this :) [21:05:25] kaldari: https://gerrit.wikimedia.org/r/#/c/214672/ i would appreciate your thoughts on this [21:05:32] MaxSem too :) [21:08:04] jdlrobson: there's a small problem, the srcset attribute is for , but the banner is a background-image because that was originally suggested by Bene [21:08:47] was there any reason bene suggested a background image? [21:10:36] jdlrobson: https://phabricator.wikimedia.org/T93106#1217820 [21:11:11] * jdlrobson looks [21:11:48] oh i see but his statement is not true [21:11:51] we can use srcset :) [21:12:23] jdlrobson: yes, probably I should write a patch migrating it to ? [21:12:54] codezee: you can do it as part of this patchset :) [21:13:01] but i've commented to explain rationale [21:14:11] jdlrobson: yes, ok :) [21:14:29] excellent [21:32:42] paging dbrant [21:34:14] JonKatz: what should happen if two merged users have collections with the same label? [21:34:19] merge/rename/drop? [21:34:40] tgr in a meeting, but also think this a question for lego? [21:35:21] it's a product question, any of those options could be implemented [21:43:47] * legoktm points back to product [21:44:37] vibha: tech support emailed me a couple days ago (just now saw it) w creative cloud invite :) [22:28:58] dr0ptp4kt: test [22:34:34] JonKatz: I'll go with rename if you have no strong feelings, since that's easier to implement [22:48:39] moushira: hey you there? [22:48:53] moushira: could you point me at the talk page you setup around re-thinking Special:UserProfile ? [22:53:25] tgr sorry for the brush off earlier, your're right that its a product question. rename sounds good to me. [23:35:01] Mhurd nice! [23:58:16] kaldari: gonna need your help merging some browser test patches in a bit