[00:20:17] (CR) Adamw: "(1 comment)" [wikimedia/fundraising/tools] - https://gerrit.wikimedia.org/r/82765 (owner: Adamw) [00:36:25] (CR) Mwalker: "(5 comments)" [wikimedia/fundraising/tools] - https://gerrit.wikimedia.org/r/82765 (owner: Adamw) [17:21:43] K4-713: mind if I sprintly-ize https://mingle.corp.wikimedia.org/projects/fundraiser_2012/cards/918 [17:21:55] Clicking... [17:22:22] One thing. [17:22:26] I need to solve the overly long fields issue [17:22:36] I like this direction better than just truncating automatically [17:22:37] I want a selector with the error code. [17:22:47] totally, thanks. [17:22:59] I was gonna put it in the data, but that's better. [17:23:24] On that note, maybe we should use a straight-up database table? [17:23:29] Thanks for both 1) doing this, and 2) mentioning it to me. That last failmail this morning made me want to tackle it myself. [17:23:35] Eh. [17:23:47] ...I'm a little... hm. [17:24:01] It's just that... the message has reached its destination. [17:24:03] That would create two inroads we'd have to maintain, rather than just the one. [17:24:19] Which is okay, if it's not the only thing we want to use it for. [17:24:48] The phase 2 task is to write a UI much like q2c/test [17:25:03] Where you can edit the message and attempt to "reinject" (process immediately) [17:25:21] aha, I see. [17:25:42] A table view of failures would make more sense than: "get next failure in queue" [17:25:48] So, my worry is that if db messages only get used when things go wrong, as opposed to constantly (like the regular queue), it's... likely to become another thing we're probably going to have to fix when we have a problem. [17:25:54] Other than just the problem. [17:27:25] OK, it's not a big deal to rework that part later. Just as long as we don't have to copy n paste from the queue management screen... [17:27:40] Yeah, that doesn't work anyway. Oh wait: copy/paste. [17:27:44] BTW, the last couple weeks of failmail are all waiting for me to do that, right? [17:27:58] hehe, yeah literal copy/baste [17:28:06] Or me. I could do that too. [17:28:06] There was one thing I wanted to do on top of this, because I noticed something. The overflowing paypal fields have redundant data in them, now. [17:28:14] whaa [17:28:20] -_2 ? [17:28:21] Check out the address field. [17:28:25] nope. [17:28:59] Yes, that's why I had high hopes for manually resolving these [17:29:08] Several of these I've seen, have city, state, and/or country duplicated in the address field. [17:29:14] Or some combination of the three. [17:29:16] but i see your point: paypal has a bad field name, maybe [17:29:21] Not even. [17:29:26] translating to "address" rather than "street" [17:29:35] It's either some change in the system, or: Right, bad UI. [17:29:43] It's not automatic, IMO [17:29:49] Who knows? [17:29:59] Taiwan (R.O.C.) is a totally antiquated thing that only a human would type [17:30:03] Anyway, I thought I might squint at some of that and... address it. [17:30:11] (pun) [17:30:20] harrrr [17:31:41] Wait. There may be a simpler way to do this. [17:32:11] awight: You were telling me the other day, about the second address line. [17:32:29] In civi. [17:32:56] Is there one that maps to what we'd think of as the second address line, instead of the first line for the second address? [17:33:07] I think we want this -damaged queue for future problems as well [17:33:07] * K4-713 goes cross-eyed [17:33:11] K4-713: yes. [17:33:15] let never speak of it, though ;) [17:33:24] No, I want to. [17:33:28] Like, now. [17:33:34] I want to let a human do that, if i read between lines correctly. [17:34:05] My plan: Explode on commas, and put as many complete sections as will fit in line 1, and chuck the rest in line 2. [17:34:09] if the address has "300 metres beyond landmark" that should be pasted, yes [17:34:12] ...if nothing is using it. [17:34:18] but I don't like arbitrarily parsing the address [17:34:33] Better than it going nowhere. [17:34:40] We can flag it for review if that makes you feel better. [17:34:42] Maybe the manual edit view has a button to do common fixes [17:35:00] I don't wanna get dogmatic on it, but review should happen before the automangle [17:35:24] There is a lot that can go wrong beyond just this field length thing. [17:35:25] There's no data loss, though. [17:35:31] K4-713: quick question -- how do I get the card view in a supertask? [17:35:50] There is one small data loss: the donor's intentionality in splitting address [17:36:13] I guess I can let go of that... [17:36:47] But... I think jvandavier is happy to fix one or two funky msgs per day [17:37:04] awight: I'd worry more about this if we cared about the address, like, at all. [17:37:10] I know [17:37:17] K4-713: ah; found it [17:37:25] * jvandavier doesn't know what he just got signed up for.... [17:37:27] hehe [17:37:32] mwalker: Sorry. But: woot. [17:37:44] jvandavier: we're discussing a way to have u correct stuff we are currently failmailing on [17:37:47] :p [17:38:28] Instead of PII barf, the failmail will link to a Civi page where you can edit the donation message [17:39:31] K4-713: ok i relent, I will automangle address first if you... just say one one more time :p [17:39:45] awight: do it. :p [17:40:41] And, yeah: I'd totally explode on commas and shove whatever doesn't fit in the first line, in the second. And, I'd be happy to get that in there if you'd focus on the damaged queue functionality. [17:40:56] * K4-713 wants to code [17:41:28] K4-713: but i thought we didn't allow you to code [17:41:29] How about we raise the street_address length, too. [17:41:42] It seems to be 64 chars? reediculous. [17:41:47] Hmm. [17:41:49] postal_code needs longering, too [17:41:49] Let me... [17:41:56] and then the automangle as well, of course [17:42:05] :p all yours, if you're serious [17:42:14] ...let me look at what it's been choking on and see if the data is ridiculous in there. [17:42:14] man i'm made of quicksilver today [17:42:42] It might take some jiggling of address_2 stuff [17:43:19] jiggling? [17:43:33] In case it was misinterpreted somewhere? [17:43:45] as address 2 line 1, rather than address 1 line 2? [17:43:45] I haven't really tested the code which allows supplemental_address_2 to actually save to the db... [17:43:57] something like that [17:44:08] or... I think it was previously chucking address 2 entirely [17:44:12] No, it's supplemental_address_1. [17:44:25] well, in the last failmail there is already stuff in supp_1 [17:44:28] The first line is street_address. [17:44:29] and it looks important [17:44:36] er. Not the one I'm looking at. [17:44:54] * K4-713 frowns [17:44:55] 'contribution_tracking_id' => '14674038', [17:44:55] ? [17:45:21] I went back to the beginning of september. [17:45:26] oh, shit josh can hear us [17:45:41] OK well the latest one is an interesting case [17:45:50] 4:45AM [17:46:37] Check '14583192'. That's one example of the more "normal" failure. [17:46:56] I loved that one. [17:47:05] It really paints a picture. [17:47:18] "Cupboard under the stairs" style. [17:47:24] we have people freaking donating from sweatshops... [17:48:04] There was one which explicitly said, "workshop" i believe [17:48:16] Yes that one's easy. [17:48:21] Dude. The one in Taiwan... [17:48:23] yes [17:48:32] it's a translation of the first address line. [17:48:35] exactly [17:48:42] * K4-713 facepalms [17:48:49] Don't tell me we need to get rid of anything! [17:48:55] There is some method to the madness... [17:48:56] Oh, hell no. [17:48:59] we just don't understand it [17:49:04] If there's stuff in there, we don't f with it. [17:49:22] This one would go in to your damaged system. [17:49:33] it sure makes for an interesting test case though. address leapfrogs over supp_1 into supp_2 ?? [17:49:34] Not... your damaged (system). [17:49:36] yeah [17:49:38] hehe [17:49:39] Your (damaged system). [17:49:51] nice algebraic grouping [17:50:00] damn hanging adjectives [17:50:10] English is Stupid. [17:50:16] dumn. [17:50:23] haha [17:50:30] Yeah, I'll... look. At these. [17:50:39] I teach my daughter some crazy speeching [17:51:07] she DOES NOT CARE about irregular orthography [17:51:23] ppsssbbbbbthttt [17:52:24] If you want to code -damaged queue as well, I can do the manual resolution UI [17:52:44] * awight jangles a few coins behind his back [17:53:25] awight: I don't know how long I have before somebody whisks me away and makes me start talking to lawyers again. [17:53:37] ok, I snatch that task [17:53:44] doit. :) [17:54:14] * awight noisly puts on a spelunking helmet [18:00:57] (PS5) Adamw: dedupe review module [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/82774 [18:46:38] awight: I don't suppose you have a guide somewhere for cloning a useful dev instance of our crm... [18:57:56] K4-713: argh. [18:58:16] 's what I thought. [18:58:30] The upgrade instructions are... pretty similar [18:58:38] Not upgrading. [18:58:40] Nuking. [18:58:42] From orbit. [18:58:47] (only way to be sure) [18:58:59] But now: Lunching. [18:59:11] k [18:59:51] Do a civi install: http://wiki.civicrm.org/confluence/display/CRMDOC/Drupal+Installation+Guide+for+CiviCRM+4.3+-+Drupal+7 [19:00:06] Then, drush en wmf_common wmf_logging wmf_unsubscribe wmf_contribution_search thank_you ganglia_reporter exchange_rates contribution_tracking paypal_audit log_audit civicrm globalcollect_audit offline2civicrm queue2civicrm wmf_civicrm wmf_refund_qc wmf_reports contribution_audit recurring recurring_globalcollect wmf_unsubscribe_qc [19:00:20] (see https://mingle.corp.wikimedia.org/projects/fundraiser_2012/cards/758 [19:01:34] * awight updates https://wikitech.wikimedia.org/wiki/Fundraising/tech/Installing_Drupal_and_CiviCRM [22:56:57] (PS1) Adamw: refactor queue functions into a new, non-singleton class "Queue" [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/85353 [22:57:02] (PS1) Adamw: WIP (FR #918) Send rejects to the -damaged queues. [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/85354 [23:05:23] (PS2) Adamw: WIP (FR #918) Send rejects to the -damaged queues. [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/85354 [23:16:04] (PS2) Adamw: refactor queue functions into a new, non-singleton class "Queue" [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/85353 [23:16:08] K4-713: Yeah, I feel like I may have accidentally lost history at some point, but I have no idea how, and I feel dumb for not having pushed it somewhere >.< [23:16:09] (PS3) Adamw: WIP (FR #918) Send rejects to the -damaged queues. [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/85354 [23:16:31] marktraceur: noooooooooes! [23:16:33] ...but, okay. [23:16:33] It would have needed a pretty massive rebase anyway, and it definitely wasn't working yet, but the DI stuff is merged and AFAIK still around in the fundraising role [23:16:43] cool. [23:16:49] I'm sure we'll revisit this someday. [23:16:53] And I think that included rsyslog [23:17:20] So you'd need drupal + civicrm + activemq if you wanted to start it up again, and to make sure the config is sane [23:17:39] I have a fantasy that someone who can set it up like second nature would be more capable of getting it working :) [23:30:26] (PS4) Adamw: WIP (FR #918) Send rejects to the -damaged queues. [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/85354