[00:11:30] ejegg: asking on the releng channel... [00:15:55] so yeah, a deployment-branch-only submodule [00:25:30] cool. I'm not overly suicidal about that :D [00:26:01] I'll go ahead and apply correction tape to the DI and crm subrepos, then, eh? [00:26:12] thanks! [00:26:42] * awight turns a color. [00:26:51] Right, I'll need to find a newer version of git for that. [00:26:59] aargh! [00:27:12] here, let me see if I can deinit those [00:27:45] k! [00:31:12] (PS1) Ejegg: Remove vendor submodule from master branch [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/195493 [00:32:32] hmm, is this going to make it a pain to switch between master and deployment? [00:32:36] (PS2) Awight: Remove vendor submodule from master branch [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/195493 (owner: Ejegg) [00:32:47] (CR) Awight: [C: 2] Remove vendor submodule from master branch [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/195493 (owner: Ejegg) [00:32:53] ejegg: definitely. [00:32:56] good :p [00:33:03] hmph [00:33:07] ok [00:33:16] that's a dirty habit... [00:33:28] that i need to train myself out of anyway. [00:33:51] what, locally testing the deploy branch before sending it out into the world ? [00:34:39] (Merged) jenkins-bot: Remove vendor submodule from master branch [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/195493 (owner: Ejegg) [00:35:42] ejegg: it's much better to have a separate clone checked out to the deploy branch [00:36:08] The especially dangerous stuff is that your development LocalSettings, for example, has dev config that will affect your testing. [00:36:19] ah, good point! [00:36:22] also: untracked files... [00:36:37] * awight counts my burn marks [00:36:46] * ejegg sets to duplicating things [00:37:05] :( [00:41:08] (PS1) Ejegg: Remove vendor submodule from dev branches [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/195494 [00:42:14] (PS2) Awight: Remove vendor submodule from dev branches [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/195494 (owner: Ejegg) [00:42:23] (CR) Awight: [C: 2] Remove vendor submodule from dev branches [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/195494 (owner: Ejegg) [00:42:47] (Merged) jenkins-bot: Remove vendor submodule from dev branches [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/195494 (owner: Ejegg) [00:42:59] *poof* [00:43:34] ejegg: aww, it's not quite that easy. because of that .gitignore abberation... [00:43:36] grr [00:43:44] I'm not quite sure yet [00:43:50] hmm? [00:44:04] but we'll have to give K4 and pizzza special instructions of some type [00:44:27] vendor doesn't go away on its own, and if you remove it, git complains that you've deleted it. [00:44:35] wait, really? [00:45:04] sorry, I hadn't pulled the merge yet [00:45:05] whew [00:45:28] Oh, I bet it's still in your .git/modules though [00:45:35] woot! it went away, at least [00:45:40] cool [00:46:21] aww, ok not sure yet. don't listen to me. I hadn't updated the submodule after removing manually. >o< [00:47:09] so, it's just: ignore the warning: unable to rmdir vendor: Directory not empty [00:47:14] then, rm -rf vendor [00:47:25] and composer install [00:47:31] yup [00:47:57] are we keeping composer.lock? Seems decentish. [00:48:25] hmm, i feel like we're somehow further away from being able to use nice things in DonationInterface CI though [00:48:35] hehehe [00:48:53] wait, how does core deal with that during CI? [00:48:55] I'm raring to just write a custom job for us [00:49:06] I think there's no composer step. [00:49:35] but how does one do something in core that depends on a new thing in vendor? [00:49:39] All we would have to do is intervene between the cloning and testing steps [00:49:42] :( [02:45:34] atgo: You still there? [02:45:44] sort of :) [02:45:51] okay [02:46:00] how are you feeling? [02:46:12] Pretty crappy. [02:46:18] But, much better. [02:46:31] Hooray for improvement. [02:46:37] woo! [02:46:40] lemons helping? [02:46:46] Oh yeah. [02:47:18] I have (or rather, had) an ice tray full of lemony goodness. [02:47:42] They have largely been mixed with bourbon and hot water by now. [02:48:07] awight isn't in today, is he? [02:48:10] Or, was he? [02:48:13] Wait. It's Monday. [02:48:21] * K4-713 sighs [02:48:26] I hate being sick. [02:48:31] And apparently drunk. [02:58:45] atgo: Anyway, I just saw a thread in which it was strongly implied that something was broken, which had a sort of energizing effect. [02:58:56] oh sorry! lost you there for a bit [02:59:00] no worries. [02:59:02] he was yeah [02:59:07] broken? which one? [02:59:17] Antifraud filter related stuff. [02:59:23] It's pretty clear to me right now what happened. [02:59:28] And everything is fine. [02:59:36] oh yeah, it's fine [02:59:42] yay. [02:59:49] Not broken. [03:03:32] * K4-713 slowly powers back down [13:52:38] Wikimedia-Fundraising, Analytics: Provide performant query access to banner show/hide numbers - https://phabricator.wikimedia.org/T90649#1104307 (Ottomata) We could make special varnishkafka instances that logged only certain requests to a topic. Not a bad idea. Do you know which of the cache clusters... [14:27:56] Wikimedia-Fundraising, Analytics, Analytics-Kanban: Provide performant query access to banner show/hide numbers - https://phabricator.wikimedia.org/T90649#1104410 (Nuria) [16:39:37] * K4-713 waves at the-wub [16:39:50] hey K4-713! feeling better? [16:40:23] I have improves sufficiently to... do things today. [16:40:27] For once. [16:40:44] Did I miss anything good? [16:41:42] It's been pretty quiet. There were some weird transactions from Afghanistan, but I think that's all sorted now [16:41:52] Oh yeah: I saw that all last night. [16:43:11] We're doing another Amazon US test tomorrow as well, in case you missed that. Same timing as last week. [16:43:20] Cool. [16:43:32] I saw awight wanted to... burn everything down, too. [16:43:36] That should be exciting. [16:43:53] heh, i'm eagerly looking forward to that [16:44:02] Oh hey, ejegg. :) [16:44:09] Yeah, that sounds like a fun one. [16:44:09] Hi K4! [16:44:16] oh yeah, that too :) [16:44:24] glad you're feeling good again [16:44:24] hey ejegg :) [16:44:34] hi the-wub ! [16:44:50] * AndyRussG waves at K4-713, the-wub and ejegg [16:45:01] Welcome back K4-713! [16:45:01] Hi AndyRussG! [16:45:02] and hello to you too AndyRussG [16:45:08] HIIIIII. [16:45:09] hi AndyRussG! [16:45:19] * the-wub just waves generally at the entire room [16:45:37] the-wub: were you just on the SafeGuard call too? [16:45:43] Just out of curiosity, does anybody have a nice neat way to check the volume of spontaneous / sidebar donations? [16:45:47] K4-713: awight and I stopped checking in vendor/ except for the deploy branch, in both DonationInterface and CRM [16:46:12] hmm, utm-medium tells us that, right? [16:46:14] no AndyRussG, I'm going to the one tomorrow [16:46:15] aye [16:46:48] the-wub: Ah right... mmm mine just ended, seems pretty decent... [16:46:50] It should also be in utm_source, and utm_... other thing. [16:47:52] I was going to run some queries if nobody already had a thing. Like in ecom, or whatever meganhernandez is using these days. [16:48:17] sorry, no dash widget yet [16:48:34] utm_medium=sidebar I think [16:48:35] pizzzacat: I have more work for you. :) [16:48:59] the-wub: It hits all three. Campaign was the one I forgot for some reason. [16:50:03] K4-713: orly? [16:50:27] Moar dash. [16:50:44] Something that would have been fun to look at today if it existed. But we'll need it again. [16:51:31] Usually, when we make news, we get an uptick in spontaneous donations. [16:51:36] K4-713: the x-by-y widget does have a 'medium' filter [16:51:51] We should sue the NSA every week! [16:52:08] ooh that's great! [16:52:19] hehe [16:58:51] We've had this idea for a while, that we can... locate news stories about us quickly, by looking at sidebar donation frequency. [16:59:07] We just need to know what's normal, and be able to identify regional spikes. [16:59:18] I think that sounds really cool [16:59:30] We've been able to do this retroactively for a while... but I want to do it frontwise. :) [16:59:42] how do you do it? google? [17:00:03] google mentions or whatever it's called [17:00:14] Nah. Just looking at donations coming in for spontaneous / sidebar donations, digging around for the country of origin, and then search for it. [17:00:52] oh ok [17:08:47] K4-713: page view spikes in general might also work, no? [17:09:01] Not my department. :p [17:09:06] Heheh [17:09:16] It's a bit like the spotted owl [17:09:17] Also, I suspect that signal to noise for us is more favorable. [17:10:09] It was a thing about old growth forest destruction. Not sure if it was the spotted owl, but it was, choose some specific species as the indicator of a change in the bigger more complex system [17:10:29] http://en.wikipedia.org/wiki/Indicator_species [17:11:11] I guess a dip in sidebar donations could also mean... negative publicity? [17:11:24] Also you'd want to filter out noise from countries with campaigns running [17:11:52] AndyRussG: That's easy, because people go through the campaign for those. [17:11:58] Not the sidebar. We can tell the difference. [17:12:08] Right [17:12:23] Social media should also show up as socialmedia. [17:12:25] But filter out the fact that there'd be no spike in that case [17:47:04] (CR) Ssmith: "Would you mind moving this to a new branch? I would like to add to it, and work with it when master is reviewed/merged to a viable state. " [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/193395 (owner: Ejegg) [17:55:47] hey sf team - FYI i'm going to try to make a habit of heading to the farmer's market at crocker galeria on Thursdays over lunch, if any of you would like to join. [18:03:52] (PS1) Pcoombe: Correct language code for Norwegian aggregation [wikimedia/fundraising/tools/DjangoBannerStats] - https://gerrit.wikimedia.org/r/195612 [18:08:42] ok, I know I usually post cats, but this dog really cracked me up today: http://i.imgur.com/FkgPC4J.jpg [18:22:32] hehehehehehhehe [18:22:49] so Honey started on a new pain medication for her arthritis this week [18:23:10] and the side effects are hilarious (apparently she's supposed ot get used to it within a week, so i'm not a jerk for laughing at her) [18:23:34] she's been totally spacey and weird [18:44:07] (CR) Ejegg: [C: 2] "This is definitely a good start. Buncha TODOs, but let's do them as a swarm of little patches" [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/190343 (owner: Ssmith) [19:08:45] qyb/imgu [19:08:51] ugh [20:28:09] (PS1) Ejegg: Whitespace [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/195655 [20:31:48] Oooo! My kids just made me a snack. It's a bowl full of peanut butter with rasberries on top. It's on a plate with pieces of bagels along the edges, the idea being that I should dip the bread in the peanut butter rasberry mélange [20:32:20] Mmm! Very peanuty! [20:32:33] (Abandoned) Ssmith: Chart now renders [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/193535 (owner: Ssmith) [20:32:45] (Abandoned) Ssmith: final x by y chart commit [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/194243 (owner: Ssmith) [21:00:58] AndyRussG: Yay Snacks! [21:01:08] AndyRussG: I feel like your update belongs in #wikimedia-snacks [21:01:14] totally [21:01:22] Oh! Didn't know there was such a thing [21:01:47] * AndyRussG eats mooooooooooooooooore peanut butter [21:01:51] oh yes, yes there is [21:01:58] shhh. [21:02:06] Don't talk about... snack club. [21:02:09] :p [21:02:37] o_O [21:40:35] ejegg: Hum. Question for you... [21:41:28] ...do you, by any chance, have any artifacts from your own hiring process? Like, the exact wording of the task that you got? [21:41:34] And if so, can you... forward that to me? [21:42:36] hi K4-713 [21:42:49] sure, I ought to have that in my gmail someplace [21:43:00] Cool. That would be really helpful. [21:43:13] I don't think the link we used before, works anymore. [21:43:28] Unless you have a different one. [21:45:04] was it a jobvite link? [21:45:10] Nah. [21:45:16] I mean, the link for the sample XML. [21:45:22] still sifting through messages - it was a long process! [21:46:30] ooh, the ~kaldari link on toolserver [21:46:37] lemme see if i copied that to github [21:46:52] ejegg: Just don't post it here if you did. :) [21:48:01] heh, oh yeah [21:49:06] dang, nope. maybe it's on my home computer [21:49:28] Ehh. I have enough to go on that I could make it again and throw it... somewher.e [21:50:14] Maybe deliberately make it slightly incompatible with the old ones. [21:51:42] hehe, nice [21:54:00] I swear I saved this before, though. [22:24:51] I think something weird happened to my database :9 [22:24:54] um [22:24:55] :( [22:25:06] not :9 [22:25:12] it doesn't make me hungry [22:25:16] just annoyed [22:25:51] everything is in the database… nothing changed in my repo… I restarted mysql…. [22:26:00] nothing perceptible has changed. [22:27:16] no errors [22:29:11] (CR) Ssmith: [C: 2] More whitespace [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/195765 (owner: Ejegg) [22:29:17] (Merged) jenkins-bot: More whitespace [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/195765 (owner: Ejegg) [23:07:29] ejegg: got a sec to let me know if u have any thoughts on this scheme? http://etherpad.wikimedia.org/p/CampaignAssociatedMixins [23:08:27] The alternative is to actually send the required JS in the choiceData and have most of the required logic in general functions in bannerController.lib [23:08:37] Mmm lemme add that to the pad... [23:13:09] AndyRussG: taking a look [23:14:31] Approach A definitely looks better! [23:14:34] ejegg: thanks! [23:14:38] Yeah I just added in B [23:14:49] A will take advantage of the whole RL infrastructure [23:18:08] AndyRussG: so we'll dynamically load them on the client, but we're adding them to wgResourceModules on the server in a traditional way? [23:18:11] ejegg: if you can think of drawbacks, possible issues? From reading the RL code it looks like the RL modules are cached indefinitely until they are deprecated [23:18:18] ejegg: yeah! [23:18:29] yeah, caching was one issue [23:18:30] and they just get updated in the cache as they are updated [23:18:45] I was thinking of asking our RL gurus about the details... :) [23:18:49] just to make sure [23:18:53] Right, so we would have incentive to re-use module names [23:18:59] hmm [23:19:18] oh, deprecation will remove them from localstorage? [23:22:10] ejegg: I think so! If I understand correctly, there's a registry of RL modules sent in with the startup module every time [23:22:43] ahh yeah. we don't want to spam that either [23:23:38] yeah... well I don't think it'll be too many modules [23:23:53] it would be interesting to look at change logs of some of the js used in the winter fundraiser [23:24:08] to see how many of the changes are just numeric parameters [23:24:39] Hmm! good point :) [23:25:03] * ejegg needs to refresh memory on how banners pass params to mixins [23:25:19] They appear there in the edit banner page [23:25:49] Don't remember where they're actually stored tho! [23:26:18] Frequent changes in JS is an issue for any move towards mix-ins in general, since code changes will now need CR and deploying [23:27:04] right, this whole effort assumes we can reduce a lot of the tweaking they do during a fundraiser to function parameters [23:27:22] indeed [23:28:02] Hmm lemme see [23:31:41] ejegg: here's a bunch of scripts in one banner https://meta.wikimedia.org/wiki/Special:CentralNoticeBanners/edit/B15_0310_enNODK_ipd_lg_twin1 [23:31:59] The small ones I think could go directly in a lib that's always loaded, I think [23:32:06] maybe [23:32:58] I guess some JS could still go in banners too [23:34:00] Recall the main features-facing reason for this shift is to move the hide logic to before the banners [23:34:22] right [23:34:41] and that's not going to depend on the myriad different element IDs that could be interacting [23:35:21] does seem to be an irreducible bit of banner-specific jquery [23:36:02] let's make all interactivity css3 animation [23:36:25] heheh another fun improvement 2 be sure... [23:40:37] Also we could optimize to see which bits of script are used often at the same time [23:42:25] yeah, that would make sense [23:43:29] Some are definitely more volatile than others [23:44:12] most of the code changes seem to be actual code, not numeric tweaks [23:44:47] could ask the-wub if there were a lot of little number tweaks during the dec campaigns [23:46:59] Ah hmmm [23:58:22] ejegg: oops sorry minor net outage 8p [23:58:54] Yeah I think it's very clear that we're in no way doing away with banner-side JS