[00:10:00] (PS4) Ejegg: WIP: add is_YYYY_donor and latest don data to wmf_donor [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/175794 [00:18:06] bahaha, just got back "Greetings Katie, Do not be alarmed." [00:18:12] Every email to me should start like that. [00:18:25] so soothing [00:18:32] Right? [00:18:38] and like an extraterrestrial [00:18:57] I was a little surprised there wasn't a "We come in peace!" somewhere in there. [00:19:17] I mean, it worked. I'm too amused to be alarmed. :) [00:20:34] K4-713: any idea why my initial db population in wmf_civicrm.install is doing zilch (not even throwing an exception), while printing out the sql and executing it via mysql cmd line populates wmf_donor just fine? [00:20:39] https://gerrit.wikimedia.org/r/175794 [00:20:54] ejegg: Hurmfpppf. [00:21:10] I mean, I could just run those scripts independently [00:21:16] Are you, by any chance, installing several times on your local? [00:21:34] If you disable the module, you have to uninstall it all the way before it will fire again. [00:21:36] yeah, but I reset the schema_version [00:21:39] drop the tables [00:21:46] and delete from civicrm_custom_* [00:21:52] I think some other table helpfully remembers for you. [00:21:55] it's recreating the tables fine [00:22:36] but i'm doing some fancy selects into a temp table, then into wmf_donor to get the initial state of the table right [00:22:55] Heck, I can even see drush running the queries in the mysql query log [00:23:15] but it's like they all get rolled back or something [00:23:26] ...not that I see a rollback [00:23:32] but wmf_donor ends up empty [00:24:44] hey there K4-713 i think you put this irc but i missed it. i’m good to put AU up in the banners? [00:26:07] meganhernandez: Yes! With bpay enabled and everything. [00:26:41] The nightly is still in a state where I have to poke it along every day, but that's a once a day thing and totally on our end. [00:33:00] bah, data truncation due to garbage in original_currency [00:33:10] plus stricter connection settings via drush [00:33:40] I don't like that "garbage in original_currency" thing you just said. [00:33:48] There's a card for it [00:33:54] ejegg: Can you tell how old the garbage is? [00:34:10] this is on my local, but there's definitely some in prod too [00:34:18] Ah. What made it? [00:34:19] and not too old in prod either [00:34:27] I'm not sure, maybe imports? [00:34:29] fooey. [00:34:45] figuring that our is the first part of the card [00:34:48] *out [00:34:52] I think we knew that manual entry was making some oddness, too. [00:35:03] Ooh, probably that too! [00:35:19] This is one of our most favorite panic buttons, actually. [00:35:26] Especially since original_* are parsed out of the 'source' [00:35:52] "OMG, SOME PROCESS IS ENTERING BAD DAToh, it's one of the people that actually logged in. nvm" [00:36:06] heh [00:36:11] That happens at least once every six months. [00:36:24] I mean, we could do some validation [00:36:36] Particularly if we're worried about extra fields that real people can't edit... [00:37:14] Yeah, we could always do extra validation, but I want to do it in such a way that we're not hacking civi core for our own mysterious purposes. [00:37:34] Those do not to well through upgrades. [00:37:37] *do [00:37:54] true. Something in a civicrm_pre hook, perhaps? [00:38:04] AU is up [00:38:09] will send to mailing list [00:38:12] I forget if those can cancel the insert tho [00:38:31] meganhernandez: Yay! [00:38:40] meganhernandez: How long? [00:38:48] just went up a couple mins ago [00:38:51] but it’s low level [00:38:57] want me to turn up to 100% [00:39:01] blast off? [00:39:33] meganhernandez: I mean, the logs look good. [00:39:53] 0:) [00:40:30] ejegg: Yeah, I think that question is why this discussion usually trails off into oblivion as soon as somebody says "Deadline over here!" or somesuch. [00:40:42] "Nice to Have" and all that. [00:40:46] heh [00:41:10] meganhernandez: I've been seeing kind of a lot of traffic all day. [00:41:20] that’s email [00:41:33] well the early morning was 100% full screen banner test (6am) [00:41:38] and then 300,000 emails [00:41:40] today is a big day [00:41:43] Aha. [00:41:45] Yes it is. [00:41:59] I'm getting a lot of mobile right now, looks like. [00:42:07] not as many emails tomorrow. but then 300,000 blast off again on monday [00:42:08] US and CA. [00:42:20] that’s interesting…we’re not running mobile in canada right now [00:42:37] Oh wait: Just kidding. Canada is email. [00:42:49] mobile is... all US. [00:42:50] yeah, that’s right [00:42:59] yeah mobile is only up in the US today [00:45:12] Yeah, it's cruising along... [00:45:21] CN seems happy, too. [00:49:46] K4-713: Just checking w/ that low-level campaign in fact [00:50:53] AndyRussG: Everything looks great from payments. [00:51:02] K4-713: fantastic [00:52:13] (PS1) Ssmith: Add amount per second chart views [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/175903 [00:54:12] K4-713: yea I am getting banners served for the campaign "C14_enWW_dsk_lw_FR", running at 40%. But they're all getting hidden with reason:waitimps, even when I delete all cookies. I guess I should assume that's intentional, I don't see anything amiss with CN itself... [00:54:46] That might be one that waits initially, until you hit a threshold. [00:55:01] meganhernandez: Does what I just said make sense? [00:55:45] yes, that’s the low level campaign running [00:55:52] AndyRussG: The idea there is that we don't want to kill people who obsessively delete their cookies, or have all of them set to delete on browser close. [00:56:41] so the banners are set to hide from each reader for 5 page views, and then show just 1 impression. 30 days later, a reader can get another banner, if they’ve seen 5 pages. [00:56:49] so that scenario is running at 40% [00:57:27] I like it when I remember things about stuff. :) [00:57:49] Anyway, that behavior is in fact expected. [00:59:32] (PS5) Ejegg: Add is_YYYY_donor and latest donation to wmf_donor [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/175794 [01:02:52] stufffff! [01:03:25] K4|brb: might be ready for some review: https://gerrit.wikimedia.org/r/175794 [01:03:34] hey AndyRussG do you want to do a 100% test with bucket A, B, C & D in the US tomorrow morning? [01:10:11] (PS1) Krinkle: Use gmmktime instead of strtotime [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/175907 [01:11:13] (CR) Ejegg: [C: -1] "Code looks good, but needs bower_modules submodule update" (1 comment) [wikimedia/fundraising/dash] - https://gerrit.wikimedia.org/r/175161 (owner: Ssmith) [01:12:48] ejegg: Might be? [01:15:25] meganhernandez: K4-713: thanks for the explanations! [01:16:04] meganhernandez: sure! I can be around as early as 13:25 UTC [01:16:56] K4-713:is ready for review! [01:17:11] ejegg: Looking. :) [01:17:38] Will put in another patch to point the new report to that table [01:18:37] (PS2) Krinkle: Use date_create instead of strtotime [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/175907 [01:18:48] ejegg: Hummm. [01:19:33] ejegg: I'm a little worried about having the db update come through the install, actually. [01:19:45] Ah, ok. [01:20:10] I think it's just that we have so very much data. [01:21:03] Hmm. Well, as long as we run it while thw QC is paused, we should be good [01:21:05] I may be... in skeptical of everything mode. [01:21:31] At the very least, though, this would be one of those things I'd want to test in a secure sandbox before we try it in prod. [01:21:33] We could even run it just for the 1k+ smart group donors [01:21:53] just for LYBUNT [01:21:57] And, if it's part of the module install, that makes the deploy totally fiddly. [01:22:14] OK, I'll break it off into another script [01:22:49] Thanks. Hopefully it's not too painful. [01:23:17] Nope, should be fine. I'll take the generated version to avoid some copypasting [01:23:27] I'll comment in gerrit just so it's on record. [01:23:30] ty [01:27:07] ejegg: Funny thing is, any other week of the year and I probably wouldn't worry about it. [01:27:32] Y'know, I like the idea of splitting it off too. [01:28:03] If only so we can decide just how much of the table we need to update for now [01:28:41] (CR) Katie Horn: [C: -1] "One inline comment about the install script" (1 comment) [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/175794 (owner: Ejegg) [01:29:10] ejegg: Yeah, that too. [01:29:22] If we try it in the sandbox and it takes forever, for instance. [01:29:53] (PS3) Krinkle: Don't use UNIX timestamp for wgNoticeOldCookieApocalypse config [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/175907 [01:30:02] All MG needs LYBUNT for is for 1k+ donors [01:30:17] ejegg: Yeah, but once they have that... [01:30:21] I'd eventually like to use this rollup for the Silverpop export too [01:30:22] ...everybody'll want one. [01:30:26] heh, yep! [01:31:04] (CR) Krinkle: "Verified that the produced number is the same, and that the setting is not used anywhere in any wikimedia/* git repository (especially not" [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/175907 (owner: Krinkle) [01:32:56] AndyRussG: ^ for later :) [01:33:53] Krinkle: Thanks for all the last-minute help. [01:34:08] (PS4) Krinkle: Don't use UNIX timestamp for wgNoticeOldCookieApocalypse config [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/175907 [01:35:06] K4-713: Anytime :) Fundraising deserves the attention. I'll try to keep an eye out more often. [01:35:29] Interesting stuff and interesting set of problems. [01:35:58] Krinkle: It is very much appreciated! We have more problems if you're looking for some... [01:37:11] (PS6) Ejegg: Add is_YYYY_donor and latest donation to wmf_donor [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/175794 [01:37:11] Can't make any promises. But I'm always looking for peacemeal things in random places I don't belong to learn more in general about different things, to spread knowledge, and to do while crunching stuff in the back of my mind. [01:37:41] Good to know! I'll put you on the list. After I make a list somewhere. [01:38:23] If you're getting stuck somewhere or need something to happen elsewhere, I'll see what I can do. [01:43:38] Krinkle: thanks! (I'm only semi-here right now but will be fully here again in 15 min or so) [01:49:08] K4-713: think I'm gonna call it a night [01:49:17] I'm about to go Science anyway. [01:49:25] So, good timing. [01:49:29] enjoy your science! [01:49:31] ?? [01:49:42] I just sent an email to Jeff (and the rest of you) about the test sandbox. [01:49:48] oh [01:49:55] Hopefully he doesn't read it and go "what?" [01:50:11] I swear we do this at least three times a year, though. [01:50:24] It shouldn't be a problem. [01:52:07] >_> [01:52:08] <_< [01:52:08] OK, well, enjoy, and have a good night. [01:52:12] Ha, you too. [13:54:32] hey AndyRussG all good for banneres to go up in a couple mins? [13:57:56] Hi meganhernandez...! No problems reported, so yes I think so :) [13:58:30] great, this campaign is set to start in 2 mins [13:58:32] https://meta.wikimedia.org/w/index.php?title=Special:CentralNotice&method=listNoticeDetail¬ice=C14_enUS_dsk_FR [13:59:43] meganhernandez: OK, looks good! [14:16:44] meganhernandez: I got the full-screen banner, but I don't get the top banner afterwards. I found the problem, also realized it's something I should have forseen: the Javascript code that moves people to the buckets C and D has to be updated. Sorry about that :( [14:17:40] Some minor changes to this javascript are needed: https://meta.wikimedia.org/wiki/MediaWiki:FR2014/Resources/Oct2014FullScreen.js [14:17:50] ah ok, yeah the banners in c & D have about 15 donations each [14:19:40] want me to take the test down & then we can try again after it’s updated? [14:20:25] that's you're decision of course... I think it'll be a quick fix... [14:22:14] ok im done with the test that’s running in a & b [14:22:18] i’ll take it down [14:22:48] it’s down. i [14:23:08] i’ll make a new test for buckets A &B. let me know when you’re ready for banners to go back up [14:24:30] meganhernandez: OK... This is something to work on with Peter... [14:24:54] ah ok, i didn’t get that [14:25:12] will you send him an email to let him know what we need? [14:26:37] meganhernandez: OK, will do that right away... [14:56:25] Hi the-wub :) [14:57:00] hey Andy :) [14:59:09] How's it going? I just sent an e-mail about changes to the JS to move people to a new bucket, really sorry I didn't think of that before! [14:59:29] ah yeah, just saw that. will get right on changing it [14:59:51] Ah cool! I think it should work, but I haven't tested it--are you able to test easily? [15:02:02] I'll make a new banner to test on. Are we using the client-side banner choice everywhere now? [15:08:28] the-wub: yep! it's activated everywhere now :) [15:09:28] So the per-campaign buckets lasting the duration of the campaign + 30 days are also enabled everywhere [15:10:20] the-wub: You can use the same method as in the e-mail to change the buckets for a campaign anytime, in fact [15:13:02] Morning atgo :) [15:20:25] hmm... it breaks when testing a banner (putting banner= in the url) since bucketsByCampaign is null then. but I think we can just skip over changing buckets in that case [15:24:32] the-wub: hmmm... yeah didn't think of a test harness for this kind of code. The only thing that comes to mind is maybe pausing in the debugger, setting a value for mw.cnBannerControllerLib.bucketsByCampaign manually, then continuing the debugger [15:26:33] you could say mw.cnBannerControllerLib.bucketsByCampaign = { name_of_campaign: { start: '1417010400', end: '1419606000', val: 0 } }; [15:26:59] I don't know what you get as the name of the campaign when you test like that... [15:28:00] the-wub: or if you want to activate it on aa.wikibooks or something... let me know if you'd like me to test it... [16:20:59] morning ejegg :) how's it going? [16:28:28] Hi AndyRussG [16:28:30] (PS1) Ejegg: Allow falsy values as custom field db defaults [wikimedia/fundraising/crm/civicrm] - https://gerrit.wikimedia.org/r/176003 [16:28:52] Just messing about with lybunt stuff. How's things with you? [16:29:30] (PS7) Ejegg: Add is_YYYY_donor and latest donation to wmf_donor [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/175794 [16:32:00] ejegg: cool... yea not bad... I was wondering if you have a sec to activate a campaign on aa.wikibooks with this banner: https://meta.wikimedia.org/wiki/Special:CentralNoticeBanners/Edit/B14_newBucketTest ? my centralnotice-admin rights are still in the post, I think [16:32:09] thanks in advance :) [16:32:29] AndyRussG: sure, one sec [16:32:43] thanks! [16:35:03] geotargeting matter? [16:35:21] also, user bucketing = ? [16:35:22] nope! [16:35:36] 2? [16:35:39] Mmm maybe set it to 4? [16:35:45] ok [16:35:48] Shouldn't really matter tho [16:35:51] Thanks! [16:37:56] i was just going to offer to do that :P [16:38:00] and here you guys go [16:38:10] Ah thanks atgo :) [16:38:46] hmm, thought I just enabled it, but not seeing the banner yet [16:38:55] cache time? [16:38:58] let me try private browsing [16:39:00] oh, or that [16:39:14] Ah, there it is! [16:39:32] Am I supposed to be rebucketed now? [16:39:43] methinks so [16:39:47] hey AndyRussG i’m around if you need any help with that test petersent [16:40:08] meganhernandez: thanks, we're just checking it out now on aa.wikibooks [16:42:28] Hey AndyRussG : idea for after the new year. If we're sending the category along with the banner choice data, we can check the hide cookie before we even request BannerLoader [16:42:58] hey guys - logging off IRC. Will be on email though [16:43:04] and gchat [16:43:09] ok atgo, safe travels [16:43:18] atgo: OK! take care ;) [16:56:17] ejegg: poof! (That's the sound of more unnecessary round trips disappearing as a result of that suggestion!) [16:56:34] also, I'm not getting rebucketed with Peter's banner... [17:12:44] (PS8) Ejegg: Add is_YYYY_donor and latest donation to wmf_donor [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/175794 [17:12:46] (PS1) Ejegg: WMF LYBUNT uses wmf_donor instead of silverpop [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/176006 [17:14:44] AndyRussG: let me know if you want me to change the campaign params. Should it have been user bucketing = 2 maybe? [17:16:47] ejegg: no... thanks the banner code is what has to change a bit still, I think [17:17:07] Hi again the-wub, just sent u an e-mail :) [17:17:41] hey, thanks. I just changed it. sorry was in a meeting [17:17:48] Hi awight! [17:18:05] the-wub: ah OK, np :) [17:18:22] * AndyRussG waves at awight [17:19:51] awight: got some lybunt-related reviewables. If you want to take a peek, here's the one to start with: https://gerrit.wikimedia.org/r/175794 [17:20:12] the hook is the part I feel needs the most scrutiny [17:21:39] Morning K4-713! [17:22:34] ejegg: Hey! How goes? [17:22:58] ejegg: awesome, headed into that storm in a few seconds. First, I'm looking at recurring GC slowwness [17:23:18] K4-713: going great [17:23:44] I split that wmf_donor initalization script off like you wanted, and added a patch to point the new report there [17:24:02] plus a tiny thing in civi core to allow falsy db defaults for custom fields [17:24:57] How did your science experiments go yesterday? [17:25:38] Hehe, they went great. :) [17:26:02] ejegg: So, you need CR, huh? [17:26:08] How's the dash coming? [17:26:37] I think pizzzacat's off today, so we may have to drive that one in. [17:26:53] Dash looks pretty good. I had a couple suggestions that didn't get it, but I can add those [17:26:59] *get in [17:27:19] Down to the short strokes, then? Neat. [17:27:32] I will totally be the CR Monster today. [17:27:49] Jeff_Green: Did you get all those insane emails I sent last night? :) [17:27:49] Rockin, some CR would be great if you've got a chance! [17:28:28] K4-713: dev_civicrm has all the data up to some time in october. think that would be good enough to test with? [17:28:36] ejegg: Yep! [17:28:42] Cool [17:29:54] ejegg: Is there actually an easy way to roll back db changes in dev civi at this point? [17:30:15] I remember that being kind of a pain, but that... was a long time ago. [17:30:23] these particular ones are not too bad - I've got a 4 line script I've been using locally [17:35:48] the-wub: sorry to bother u (I'm around later too if you're busy), we're still talking about B14_newBucketTest? It looks to me like it's still checking mw.centralNotice.bannerData.campaign (inline