[00:03:47] is wikitech a public wiki? [00:03:59] pizzzacat: yep [00:03:59] yep [00:04:04] heh [00:04:08] bam [00:04:12] I've got two yep's [00:04:17] yepyep [00:04:26] it's like those puppets on sesame street [00:04:30] This may be a geometric problem. [00:04:49] I do try to hide behind my lower lip every time I hear a phone ring. [00:04:52] hey ejegg - just to confirm the new WP thing works for all transactions through them? [00:04:54] You may be on to something. [00:04:55] https://www.youtube.com/watch?v=cCri-vLgSNA [00:05:17] atgo: I think so [00:05:23] ok i'll try it :) [00:05:36] they say the amount of text displayed on the statement depends on the bank [00:05:41] we may need to abbreviate [00:06:11] Hurmphhblph. [00:06:32] obviously my donation failed [00:07:32] pizzzacat: Actually... I feel more like we're these dudes: http://muppet.wikia.com/wiki/Honkers [00:07:59] There was that amazing bit with the baby one that was trying to figure out where his nose was. [00:08:37] haha aww I remember those [00:13:45] awight: do you recall - was the thing we weren't happy about from GR the zendesk API interaction? [00:13:54] that we'd built for BTC [00:16:11] atgo: that was a really small project, to... [00:18:13] too [01:18:42] (PS8) Awight: WIP factor controllers, take 2 [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/185599 [01:19:07] (CR) jenkins-bot: [V: -1] WIP factor controllers, take 2 [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/185599 (owner: Awight) [01:20:02] awight_: Bah, my concentration is broken. I'm going home. [01:21:00] K4-713: off you go, then! [01:21:04] yep [01:21:06] yep [01:21:12] somehow, a Malcolm McDowell interview is helping me [01:21:14] yepyepyepyepyep, uh-huh. [01:21:19] huh [01:21:20] of all things. [01:21:22] radio [01:52:41] (PS9) Awight: WIP factor controllers, take 2 [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/185599 [01:53:05] (CR) jenkins-bot: [V: -1] WIP factor controllers, take 2 [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/185599 (owner: Awight) [14:33:39] (Abandoned) AndyRussG: Client side: don't send unused banner variance params [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188145 (owner: Awight) [15:13:28] (CR) Awight: "I think this was abandoned accidentally?" [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188145 (owner: Awight) [15:16:49] (CR) AndyRussG: "Oh oops, yeah, I abandoned the wrong one!! Sorry! 8p Hold on though, it was gonna get -1 anyway ;) explanation coming..." [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188145 (owner: Awight) [15:32:46] (CR) AndyRussG: [C: -2] "Thanks a ton for doing this, very truly appreciated!!! A valiant effort! :D However..." [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188131 (owner: Awight) [15:33:13] ejegg|away: just thinking, did we ever make a card about using Composer for Minfraud, and the CI issues we encountered? [15:33:21] I don't think I did [15:34:04] awight: Hi, sorry for my stickliness, thanks again doing the patches... :) [15:36:13] I'm sure I would have gone exactly the same route at first, BTW, so it did save a huge amount of time that you did that :) [15:36:42] (CR) Awight: "Good points, I'll respond in more detail tomorrow, or you can continue w/o me. The only thing I want to be emphatic about at this point i" [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188131 (owner: Awight) [15:37:27] AndyRussG: yeah, it was pretty deep down in the heap... [15:38:41] Heh, this is all fun new gitexperience for me, never had to do such historydiving [15:38:44] fyi once my kid is awake (soon), I won't have time to do anything else today, so you should just ignore my patches and forge on w/o me [15:39:56] It does look like the damage to https://gerrit.wikimedia.org/r/#/c/182074 should be easy to fix [15:40:01] trying that now... [15:40:41] awight: would you be OK with the feature branch idea? [15:41:11] Basically the justification is that we can't rewrite master branch history.... [15:41:56] ...but we would need to to get the cleanest possible deploys, if we wanted to only ever merge from master [15:42:22] I'm pretty sure that's a bad idea, I started to explain in the comment. [15:42:41] Hmm [15:42:41] Feature branch itself is fine, but you should merge that to master, then to wmf_deploy [15:43:10] otherwise, it's gonna burn for all futurity when you try to merge from master->wmf_deploy [15:43:56] Hmm... I thought that by just doing a removal of params, then reverting exactly that, we'd get wmf_deploy to the exact same state it's in now [15:44:07] And the we'd be able to cleanly merge master in [15:44:35] But I may be wrong, not too sure about the logic of what git would call a conflict and what it wouldn't in those circumstances [15:45:24] AndyRussG: that sounds much better than my plan [15:45:40] however, do that work on master [15:46:16] (Restored) Awight: Client side: don't send unused banner variance params [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188145 (owner: Awight) [15:46:22] (PS2) Awight: Client side: don't send unused banner variance params [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188145 [15:46:28] lemme try that [15:47:03] So wmf_deploy history would read like this (bottom-to-top): [15:47:03] - Merge in master [15:47:03] - Revert temporary removal of requirements for params (do not deploy) [15:47:03] - Temporary removal of requirements for params (for temporary deploy fix) [15:48:00] My issue with working on master is that an actual revert brings in all that old cruft, but what if we're forgetting to bring in other cruft that the returning cruft needs? [15:48:21] I'd sooner add a new patch to master that sends in dummy params to Special:BannerLoader [15:48:23] I understand the wmf_deploy plan, but if you change those lines in master ever again, git will complain [15:48:34] yes !-2 what u said [15:48:54] that's much more sane than my crazy tripartite thing [15:49:12] K let's do that :) NE way, crazy = fun! [15:50:26] but... http://knowyourmeme.com/photos/439891-the-internet-is-serious-business [15:54:01] (PS1) Awight: Back out client side "don't send unused banner variance params" changes [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188378 [15:54:04] yuck! [15:54:51] (Abandoned) Awight: Revert "Remove unused banner variance parameters" [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188131 (owner: Awight) [15:55:03] (Abandoned) Awight: Server side: don't require unused banner variance params [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188144 (owner: Awight) [15:55:09] (Abandoned) Awight: Client side: don't send unused banner variance params [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188145 (owner: Awight) [15:55:40] * awight taps fingers waiting for Jenkins to do my job... [16:01:15] AndyRussG: btw, you can test whether I'm full of it by merging that partial reversion above to wmf_deploy locally, instead of master, then beat on the relevant lines in master and try merging back to wmf_deploy... [16:02:05] Mmmm I trust you! (though that also sounds fun) [17:37:34] Woah, that was terrible. [17:38:49] ejegg: Have you ever messed with the paypal audit code? [17:39:24] I ask, because I just blew up 17 times. [17:39:54] To clarify: I think they've gone funny. Not us. [17:40:49] Yesterday we blew up nine times. [17:42:55] eek [17:43:21] nope, but I'm looking now [17:43:35] Looks like awight was on to something with ccogdill, but she's on vacation right now. [17:47:39] (CR) Ejegg: "recheck" [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/187644 (owner: Ejegg) [17:47:54] ugh, are we doing that again? [17:48:22] npm failing for gosh knows what reason [18:05:16] (PS1) Ejegg: Sample banner impressions client side [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188394 (https://phabricator.wikimedia.org/T45250) [18:13:19] ejegg: Woah, my inbox just caught up with all the actual fails. [18:13:34] Seems to have jumped up to 59. [18:13:46] Well, that's cool. [18:14:03] ++order of magnitude [18:14:11] ...from what it was doing before. [18:15:08] (CR) AndyRussG: [C: 2] "Thanks! Verified that this works when installed on a subscribing wiki with the infrastructure wiki at the current version of the wmf_deplo" [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188378 (owner: Awight) [18:15:44] (Merged) jenkins-bot: Back out client side "don't send unused banner variance params" changes [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/188378 (owner: Awight) [18:17:16] and a few impact notices [18:18:17] ejegg: It sort of looks like for some reason, one-time donations are coming in differently such that they're now hitting a recurring codepath, which chokes when there isn't a subscription? Maybe? [18:18:30] ah, that could be it! [18:20:02] It seems like the most likely thing. [18:20:36] They decide to normalize some bit of reporting on their end, which we were depending on being abnormal. [18:32:08] Samurai Cop: https://www.youtube.com/watch?v=R6OoZNdAQuA [18:32:27] pizzzacat: The one I'm advocating: http://en.wikipedia.org/wiki/The_Creature_Wasn%27t_Nice [18:32:34] (That will be the next Bad Movie Night movie for all of you guys wondering) [18:33:48] if it's got leslie nielsen it can't be thaaaat bad [18:34:26] You'd think that. [18:34:44] I have a copy. [18:34:48] the number of mullets in samurai cop [18:34:50] OMG [18:35:08] K4-713: you neeeeeed to watch (with sound) the samurai cop trailer [18:35:22] That is what pizzzacat just said IRL. [18:35:24] :56 and 1:20 are particularly charming moments [18:35:30] okay, okay... [18:35:31] :) [18:35:54] I am seriously so excited you guys [18:36:01] K4 is lolzing [18:36:25] Monte suggested it [18:36:30] soo good. [18:38:12] yeah it's incredible [18:40:31] here is the incredible "The GingerDEAD Man" starring Gary Bussey: https://www.youtube.com/watch?v=xsOtcEOkJ-I [18:40:46] "A new kind of cookie" [18:40:49] eep [18:41:08] That spaceship movie, though? That one got my brother and myself fired by the family, from being able to pick rentals from the video store when we were kids. [18:41:14] Years, that lasted. [18:41:22] haha I'd say that was kinda fair [18:41:29] We practically fired ourselves. [18:41:47] ...and then I bought it. [18:42:26] well that solves that problem [18:49:44] Wikimedia-Fundraising: Enable HSTS on https://payments.wikimedia.org - https://phabricator.wikimedia.org/T88199#1012171 (Jgreen) for nginx: add_header Strict-Transport-Security max-age=31536000; [19:02:46] (PS2) Ssmith: Add generic board functionality & update sql [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/187628 [19:04:25] (CR) jenkins-bot: [V: -1] Add generic board functionality & update sql [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/187628 (owner: Ssmith) [19:10:38] oh my god this gingerbread thing [19:10:41] gingerdead* [19:10:52] those fx [19:13:55] MediaWiki-extensions-CentralNotice, Wikimedia-Fundraising: Special:RecordImpression should die in a fire - https://phabricator.wikimedia.org/T45250#1012284 (ellery) To reiterate, to do AB-testing with sampled impressions, you need to know how many of the sampled impressions did not lead to a donation. This pa... [19:28:04] Wikimedia-Fundraising: Enable HSTS on https://payments.wikimedia.org - https://phabricator.wikimedia.org/T88199#1012339 (Jgreen) Open>Resolved a:Jgreen this is done. we start at 30 days [19:50:48] (CR) Ejegg: "I think we can simplify some more. Comments inline." (7 comments) [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/187628 (owner: Ssmith) [20:56:56] K4-713: it looks like our code is doing the right thing with the paypal audit - only doing the normalize_recurring when there event code is "T0002", for Subscription Payment [20:57:20] That's interesting. [20:57:46] So... why is it choking up over the subscription id, I wonder? [20:58:04] all the ones before today with that event code had sub ids, but those 59 in this morning's file didn't [20:58:18] ok, all 3 of the earlier ones I checked :P [20:58:21] And they were actually recurring? [20:58:28] Awesome. [20:58:55] Ahh, let me look in payments logs to see if they should have been [21:32:51] K4-713: 55 of the 59 T0002s missing subscription ids yesterday had donations with subscription ids last month on the 2nd [21:45:13] ejegg: Well, that seems fishy. [21:45:41] Sounds very much like the file got corrupted. [21:45:52] yeah, looking a lot like a Paypal-side problem [21:45:56] okay. [21:46:09] The transactions with those missing are pretty bunchy too [21:46:10] We should ask them what the heck happened, then. [21:46:44] just trying to figure out who I should pester via email [21:47:03] I was about to ask if you would write up a draft with specific transaction IDs they can investigate... but you can send straight to them, too. [21:47:08] I'll look for the right contact. [21:47:19] OK, cool, I'll send you the data [21:47:59] Oh, wait a second... this may have happened before. [21:48:18] Search your inbox for "Incident: 141211-000393" [21:48:25] ejegg: ^^ [21:48:39] I definitely see a bunch of bad_audit_line emails [21:49:02] Yeah, not that. [21:49:05] ok, I see the thread from this summer [21:49:42] duplicates! [21:50:09] lemme see if the addresses do appear twice [21:51:53] they sure don't look like duplicates. none of the emails on those 59 lines appears on any other line in yesterday's or the prior day's audit [21:52:29] and 55 of 'em had recurring donations exactly a month ago with an intact subscription id [21:53:25] wth [21:53:54] Does that make sense? [21:54:03] * K4-713 headscratches [21:54:29] looks to me like these were correctly charged (at least 55 of em) [21:54:36] and just not marked correctly in the audit [21:54:57] wonder if it had to do with all the paypal system issue notifications from yesterday [21:55:51] atgo: Are you here? [21:56:00] yar [21:56:31] I've been down a depressing rabbit hole, and I'm not entirely sure what to do with it. [21:56:54] oh dear [21:56:54] You should probably get the full report. [21:56:58] ok.. i'm in meetings until 3 [21:57:00] ...before I do anything weird. [21:57:03] haha [21:57:11] okay, I will try not to do anything weird before then. [21:57:23] sure, better to talk IRL? [21:57:30] yep [21:57:38] k [22:40:21] K4-713: yo [22:40:29] talk in like 3? [22:41:57] atgo: Sure. [22:48:03] nice hat! [23:18:55] hey K4-713 if i'm looking at fraud data, better in fredge or civicrm? getting aggregate per country [23:22:44] atgo: Er... how are you looking at fraud data in civicrm? [23:22:51] yeah i figured out my mistake [23:23:03] You should just say "magic" next time. [23:23:11] touche