[00:01:37] awight: right... well OK by text [00:02:05] maybe the cache wouldn't go caput? I mean, we'd just have to limit the number of active campaigns and bucket params being sent globally at any one time, no? [00:02:23] Which is different from now--now everyone's in a bucket and a bucket param is always sent... [00:02:53] But if only text-based bucket params are sent for active campaigns, there'd still be enough repetition to ensure a lot of cache hits, no? [00:03:12] (back shortly...) [00:03:16] I wonder if we could do something on the cache, like only cache the redirect, but serve the redirect content internally... [00:03:41] Hmm, there was also that edge somethingorother for caches [00:04:37] (CR) Katie Horn: [C: 2] Nastiness to circumvent a default setting [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/169268 (owner: Awight) [00:04:50] (CR) Katie Horn: [V: 2] Nastiness to circumvent a default setting [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/169268 (owner: Awight) [00:04:59] K4-713: thanks! [00:05:03] yep! [00:05:55] (PS1) Awight: Merge remote-tracking branch 'origin/master' into HEAD [wikimedia/fundraising/crm] (deployment) - https://gerrit.wikimedia.org/r/169274 [00:06:04] (CR) Awight: [C: 2] Merge remote-tracking branch 'origin/master' into HEAD [wikimedia/fundraising/crm] (deployment) - https://gerrit.wikimedia.org/r/169274 (owner: Awight) [00:07:39] !log updated crm from 9bb50403616d80aa8d39a89ab59965f53e9e3f3d to ffa543cab3eb508fa38b94c6de2643d168b0d507 [00:07:46] Logged the message, Master [00:08:22] That was fast. [00:09:53] ccogdill: encountering some turbulence... will keep u posted [00:10:13] okay [00:10:21] I just got a surprising "not authorised" for RGC trxn_id 8867161196 [00:11:00] hmm [00:11:15] it gives one of GC’s less helpful failure reasons in their console [00:11:35] although I think this one usually means a bank rejection [00:11:43] that does give me some hope [00:11:43] so maybe that wasn’t you awight? [00:12:05] no... unfortunately, it just became really hard to force my own contribution to go through, cos we hit midnight UTC [00:12:46] hrmm another not authorised: 416670594 [00:13:09] hmmmm [00:13:24] that one’s rejection reason is weirder, but I would still think is a bank/account thing [00:14:02] am I allowed to paste that sort of thing in here? [00:14:04] The thing that scares me, is that the change I made affects how the charges are processed, so I can imagine I might be screwing up some detail of the request. [00:14:10] ccogdill: nah, PM though? [00:23:55] ccogdill: omg, this whole time it's been succeeding. I think. [00:24:12] (back :) [00:24:16] It tries one successful charge, then retries one previously failed subscription. [00:24:33] ccogdill: could you look up trxn 757711015 [00:24:34] haha what. [00:24:50] awight: sorry to foist such complicated multitasking upon you... So basically our main barrier so far is the cache [00:25:02] AndyRussG: yes enemy and friend #1 [00:25:09] my best fiend [00:25:09] frenemy [00:25:12] :) [00:25:16] If we can make it not explode [00:25:18] that one looks good, Adam. I don’t see a failed attempt though [00:25:19] new db name / band name [00:25:22] woot! [00:25:27] ccogdill: that's great [00:25:32] seems like it! [00:25:36] well... I'll open the flood gates, then [00:25:36] And make sure we have a fast enough response on banner fetching [00:25:46] okay, I’ll keep an eye out [00:25:56] The rest is all pork rinds with green sauce [00:26:16] AndyRussG: something to think about... we used to have an extra round trip in there, we would ask the server for a list of banners, then do the selection on the client side. [00:26:32] round trips = slow. but this would have given us all kinds of wiggle room. [00:27:11] !log reenabling recurring GlobalCollect job [00:27:16] Logged the message, Master [00:28:06] ccogdill: so... it seems to work, thanks for all the console help! We should definitely be on high alert for any abnormalities in subscriptions, though. [00:28:25] definitely. I think I’ll do the query thing anyway tomorrow [00:28:38] but will let you know if the donors go crazy, as well [00:28:57] omg. everything is "-21" [00:29:05] awight: humm! That might work if we could cut out the round trip by sending the list of banners with the initial response... [00:30:02] AndyRussG: which initial response? bundled with the controller, or something? [00:30:05] * awight cries about ESI [00:31:24] ccogdill: whew. ok nvm my "-21" panic. time for homing. [00:31:36] haha okay [00:31:41] I’m fine with that [00:32:41] seems like they are ordered by subscription begin date or something... [00:32:55] I should give an explicit order, one day... [00:38:19] awight: yeah the initial http response bundled with the controller... I guess that'd be hitting php for every request, tho, so no... [00:39:31] AndyRussG: so, I think we can make the cache thing work for us rather than against. [00:39:38] All it does is enforce determinism... [00:39:42] which is a property we want. [00:40:05] We just have to find a sane way to pass URL params that contain all the information we need to make a choice [00:40:25] while ideally reducing the number of possible input states... [00:41:22] or, find a way to compress the explosion we cause at the cache layer, like the redirect thing I was daydreaming about a minute ago :-/ [00:42:09] K4-713: ejegg|away: fyi, CRM has been deployed and the new code seems stable. [00:42:19] Today's RGC job ran, the results are normal. [00:42:20] Yay! [00:42:24] Thanks for all the help! [00:42:33] That went pretty well. [00:42:37] :-/ [00:42:39] Move cards? [00:42:56] oh :) yeah I guess this means work out of Mingle again [00:43:07] * K4-713 squints [00:43:26] * awight makes that a retroactive "4" [00:43:29] I feel like we're having some kind of tool rebellion. [00:43:38] awight: OK yeah I think that's a veyr sane approac [00:43:41] hey call yourself a tool :p [00:44:27] * K4-713 tucks shirt in [00:44:31] kbye, comrades [00:49:32] hmmm [00:55:05] I was going to say more about the tool rebellion, but he's gone now. [00:55:20] It just makes me nervous when nobody is, you know... using what we have until the last minute. [00:55:30] gerrit, mingle... [00:55:36] I'm sure there are others. [00:56:39] I have occasionally caught people recreating mingle views in trello. [00:56:42] :[ [01:01:07] K4-713: soon time to do that in Phabricator [01:01:37] Let's migrate it all in phabricator and then I'll copy and paste it into a google spreadsheet we can all manipulate. [01:01:43] 8[ [01:02:17] noooooooo let's create one single tool that's a combination of all of them [01:02:25] A couple jobs ago, there was a mechanical engineer who insisted on doing all his work in excel. [01:02:33] ...this included creating mechanical drawings. [01:02:42] I didn't even know you could do that. [01:02:53] we can call it Phabriminglellohoogle Docs [01:03:14] Sorry Phabriminglellohrrittoogle Docs [01:03:22] ...better. [01:03:31] Oh wait bugzilla [01:03:33] But, where's the "zilla"? [01:03:35] ha [01:03:39] Phabriminglellohrrittoogle Docszila [01:03:54] Phabriminglellohrrittoogle Docszilanode (for Freenode) [01:04:18] * K4-713 buys the domain name [01:04:43] There's a gold mine in waiting! [01:05:29] I'm pretty sure you're right. [01:05:38] Mostly, though, I just want to get my email there. [01:05:40] Creating technical drawings in Excel is only a little harder than doing it in Latex [01:05:44] heheheh [01:06:08] He was probably a LaTeX table refugee [01:06:18] He did have a glorious beard. [01:06:29] The proof of the pudding [01:06:39] The kind of beard nobody argues with. [01:07:43] Yow. Did you happen to see our team velocity graph recently? [01:07:46] https://wikimedia.mingle.thoughtworks.com/projects/online_fundraiser/overview [01:07:58] Scroll down to "All Tech – Sprint Velocity". [01:08:49] Is that good or bad? [01:09:30] The sprint we're on, destroying all the other ones? [01:09:57] It's... Somebody released the kraken or something. [01:10:08] We're gettin' kraken [01:10:31] Seriously though. [01:10:32] Wow. [01:10:41] I think we finally deployed all the things. [01:10:47] Like, really: All of the things. [01:10:53] Yeah that did happen [01:11:09] I just did a very very teeny patch that was worth "1" [01:11:19] It took me all of 20 minutes to write and test locally [01:11:26] And we don't have a new sprint until wednesday. [01:11:39] We have... one day to get one more point. [01:11:43] Which I think we have coming anyway. [01:11:56] So a lot less than 1/2 of the 2 points worth of time I've spent already on GlobalAllocations memory stuff [01:12:07] heh [01:14:12] uh-oh. We may already be at 50 points. [01:14:22] Oh there it is [01:14:43] I pressed control-W with the wrong window activated and the channel was gone, poof! [01:15:13] I gotta be careful with this window manager, one false move of the mouse and you're control-whatevering on the wrong window! [01:24:27] We also need to make a card for current work on a new bucket proposal, I guess [01:33:13] Oh man. [01:33:21] You guys have had a lot of meetings about buckets lately. [01:33:26] ...how's that going? [01:33:37] Not too bad [01:33:57] I think [01:34:18] I feel pretty good about the requirements we got, for the most immediate fix, from today's meeting [01:34:49] Oh, good. [01:35:03] There were some fairly complicated ideas that came out of an analytics chat this morning. [01:35:16] Cleared up some misconceptions (for example Ellery's use of "test", vs. "experiment") [01:35:22] Ah? What happened there? [01:35:26] I'm... having a hard time getting over the fact that it's November in a week. [01:35:40] 4 days actually [01:35:43] A lot of these things will be cool to explore. [01:35:46] ...for next year. [01:35:53] Or maybe other localities. [01:36:03] Other localities that ramp up to next year. [01:36:17] Ah hmmm [01:36:38] I'm particularly wary of things that have the potential to increase latency. [01:36:47] Analytics always brings a lot to the table, it's fun to have them as one of the stakeholders for CentralNotice [01:36:53] Oh yes. [01:37:24] Yeah totally can't ignore that issue anywhere [01:37:38] I'm also a little worried about how we're thinking about CN at a vehicle for us to do tests being a priority over it being a vehicle to, you know... notify people about things. [01:38:05] Hmmm [01:38:07] Yes [01:38:21] Well if I had my druthers it would all be separated out into independent stuff [01:38:34] One system for organizing tests [01:38:43] Including fundraising and usability and features [01:38:54] Well... lots of people want tests. [01:39:03] And not just for banners. [01:39:11] Another for allocating features (including banners, but also whatever) based on location, project, device, etc [01:39:27] These concepts are currently married in a lot of people's minds, and that's probably not great. [01:39:31] Another for injecting features on-the-fly (as we do with banners, but there could be more stuff) [01:40:11] I think we could do something less crazy for other page elements. [01:40:39] Well the thing is there are lots of posibilities for making kinda mashupy things [01:40:50] I don't think people want that, though. [01:40:54] Muddies all the tests. [01:41:08] or maybe they don't yet know that they want it [01:41:30] Have I shown you my most recent favorite programming book? http://books.google.ca/books?id=xColAAPGubgC&printsec=frontcover#v=onepage&q&f=false [01:41:43] http://en.wikipedia.org/wiki/Domain-driven_design [01:41:50] Well, we don't want to run into the situation where people don't know if people actually like the thing they did, or if it's getting some booster from a subset of people getting a different test as well. [01:41:57] Heh, no. [01:43:18] Mmm a few places I've been saying that in CN the domain model is wrong [01:43:41] This is where I get the concept [01:44:15] It's like, long-term, I would want to scope out aaaaall the requirements and get a better domain model [01:44:45] Which is just the software's model of what the software is about in the real world, in terms of the concepts they actually use... [01:45:07] boiled down and simplified into something formalizable and codeable in software [01:45:32] Heh. Good luck with that. :) [01:45:46] Hmmm yes that was kinda Adam's reaction too 8p [01:46:00] I think the problem we have with CN, is an ongoing scope creep over the course of... [01:46:15] Yes that too [01:46:16] ...at least four years? Five? [01:46:35] So, the initial assumptions were almost immediately invalidated. [01:46:42] WRT domain model, a lot of CN is targetting. Really targetting should include project, language, device, bucket. But device is placed on banners while most other targetting is on campaigns [01:46:43] Some of them are probably still in there. [01:47:29] I think I know why that was on the banner level. [01:48:06] If you think of a banner in the, ah... "I should tell all the people about [thing]" sense, it's completely sensible. [01:48:21] Ah yes [01:48:22] Because you'd have a version of that message that worked for all the people. [01:48:32] Yup [01:48:53] This is why I want somebody not in fundraising to work on this thing. [01:49:08] You know, with us. [01:49:17] Yeah [01:49:25] I do understand :) [01:49:34] We are entirely too prone to scratching our own backs. [01:49:55] Hmmm [01:50:15] I mean, we have to be. [01:50:21] It all makes... total sense. [01:50:47] A danger in many different contexts, also from the perspective of the foundation as a whole [01:51:14] I think this is what you get when you are bare-bones resourced for a while. Just, you know: In general. [01:51:39] Greedy algorithms for scheduling new work. [01:51:59] Hum right that makes sense [01:53:02] I guess we're still really so bare-bones it'd take about an infinity of growth to become not so [01:53:22] It would be a while. [01:53:30] And it always gets worse before it gets better. [01:53:43] * K4-713 grins like a maniac [01:53:59] (I'm actually kind of good at that part, though. I have proof) [01:54:11] Hmmm you mean when new people come on board and take everyone's time while they learn the ropes? [01:54:20] Oh yes. [01:54:27] Also, they are different people than we have ever had. [01:54:30] Fresh ideas. [01:54:34] Not... predictable. [01:54:52] Heh well that part sounds good [01:54:56] I mean, maybe at first. But then they get comfortable and knock everything sideways. [01:55:11] Hmmmm [01:55:25] And if I've done my job at that point, it's knocked sideways in a way that's, you know... beneficial. [01:55:27] Though... what if everything was just supposed to go sideways? [01:55:34] Exactly! [01:55:45] Usually things are more stable once they have been knocked over. :p [01:56:04] Heh tests pass when all tests fail [01:56:35] I dunno [01:56:45] At the very least there isn't much prospect of it being boring [01:56:52] Truth. [01:57:44] Just so long as it's not so stressful as to pull people down (which I guess it sometimes is) I think it's fine [01:59:06] Also stressful for a lot of people in a very different way than other jobs, since everyone's so invested in the project in itself [01:59:34] Well, that's where it usually goes all to hell. The... stress. But that can be dealt with. [02:00:28] Aaaaanyway: fun times. [02:00:37] At least we're not bored! [02:00:42] That's the worst. [02:03:35] * AndyRussG eats 2-week-old chocolate chip cookie fragments to promote mental concentration :) [02:05:35] I... should go and eat actual food or something. [02:06:11] So, I will! Have a good one; I'm outta here. [04:58:32] (CR) Ejegg: [C: 2] Make state field wider, zip smaller to compensate [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/169230 (owner: Ssmith) [04:58:50] (Merged) jenkins-bot: Make state field wider, zip smaller to compensate [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/169230 (owner: Ssmith) [10:43:03] (PS1) Ssmith: Add form partials for WP button copy [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/169341 [12:07:09] hi jessicarobell, quick question about the worldpay form buttons! [12:07:25] hi pizzacat! [12:07:29] sure [12:07:48] so I added the new button copy tests, and now am working on the button colors [12:08:12] do you want each color (the current color, grey, and blue) for each version of the button copy, or just the current copy? [12:08:12] great [12:08:29] so like, [12:08:38] soutenez-nous in grey and blue [12:08:45] soutenez wikipedia in grey and blue [12:08:46] etc [12:09:01] or just the current copy (continuer) in grey and blue [12:09:29] I assume the current copy, but wanted to check [12:15:54] Good question! How difficult/time consuming is it for you to set up these forms? The reason why I ask is that we don't know the results of these tests yet. Is it easier for you if we just get back to you as we are running the tests, to ask you to set up an extra one once we have the results of the first test? or is it better for you to just set up all variations right away? [12:17:49] it's not that difficult to set them up. as long as I have clearly defined designs it's pretty quick. [12:18:26] does this mean just do the blue and grey versions of "continuer"? [12:18:31] (for now) [12:18:41] and then the others later, depending on how they do in the tests? [12:19:05] also, please give me at least one business day to make room for them :) that's all! [12:33:14] jessicarobell is that ok if I go ahead with the colors on "continuer" only? [12:37:02] Ok that sounds good pizzacat. Thank you! :) [12:38:03] no problem! [12:52:10] hope you are still enjoying Berlin! When are you leaving? [12:59:19] jessicarobell I'm leaving Thursday! [12:59:25] I love it. I will miss it. :) [13:04:49] ah. good to hear. Enjoy your last few days! [13:05:02] Have some K-fetisch! ;) [13:30:12] (PS2) Ssmith: Add form partials for WP button copy [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/169341 [13:43:16] pizzacat: you still there? :) [13:52:28] (PS3) Ssmith: Add form partials for WP button copy [extensions/DonationInterface] - https://gerrit.wikimedia.org/r/169341 [15:51:01] (PS1) Awight: Wean remaining usages of legacy DonationInterface checkout [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/169510 [15:57:39] awight|virtumeet: https://docs.google.com/a/wikimedia.org/document/d/150lrUt8b4aiPK19XCcoE6Yy3TwN_wo529UGQjk0eTbo/edit [15:57:45] also: hi! :) [16:01:05] awight|virtumeet: one sec... [16:03:06] AndyRussG: k! [16:03:10] ejegg: ping... [16:03:19] https://plus.google.com/hangouts/_/wikimedia.org/buckets-and?authuser=0&hceid=YXdpZ2h0QHdpa2ltZWRpYS5vcmc.qeco0r73bduti9cdbperiqvm88 [16:03:32] hello [16:23:41] * K4-713 waves at pizzzacat [16:24:07] hiiiiii K4-713 [16:24:48] ...I gave you my localsettings file, didn't I? [16:24:57] yes [16:25:20] I knew that, because your apache logs indicate you're looking for the "logo" image I'm using. [16:25:35] Oops. :) [16:25:53] haa yeah, it looks like that's it's only complaint, from what I can see [16:26:23] If you... search your localsettings file for that reference and just comment out that line (or replace it with something that amuses you), that will stop complaining. [16:27:18] But! This isn't the thing we're looking for anyway. [16:27:46] ohh k [16:28:39] Okay: Now we're going to play 20 questions. [16:28:53] Did you manually do this setup, or is it in vagrant? [16:30:09] manually. [16:30:18] * K4-713 looks relieved [16:30:21] this is the one you tipped me off to the symlinking workflow with [16:30:29] :) [16:30:36] I have never actually used vagrant myself, so I would have had to fake the rest. [16:30:42] Er. What OS are you running? [16:30:58] osx [16:31:05] ...d'oh. [16:31:15] :-D [16:31:18] haha [16:31:22] Okay, I'm going to need a minute. [16:31:27] har [16:31:42] no worries. [16:31:45] I remember this was a PITA compared with how it works in ubuntu. [16:31:59] lame. [16:32:54] Try... [16:33:02] grep worldpay /var/log/system.log [16:33:20] from within the fr instance? or anywhere? [16:33:24] command line [16:33:34] but like the dir doesn't matter? [16:33:36] And, if you're not using vagrant, not in there. [16:33:56] Nah. Anywhere is good. [16:33:56] ok [16:34:01] done [16:34:04] Get anything? [16:34:10] yes [16:34:13] WOOT [16:34:18] :D [16:34:24] want me to pastebin? [16:34:28] Sure [16:34:42] yeah this looks way more relevant [16:35:13] https://tools.wmflabs.org/paste/view/025b5e3e [16:35:28] Try this too: This is fun. pop open a shell and run: [16:35:39] tail -f /var/log/system.log | grep gateway [16:35:48] Then leave that open and refresh the page you were looking at. [16:37:37] hmm [16:37:41] it didn't show anything [16:38:09] What were you looking at? [16:38:36] It probably won't do anything until you hard reload a payments form. [16:38:43] I refreshed the form in the browser and then went back to the shell [16:39:16] Curious. [16:39:19] hmm yeah nothin' [16:41:07] Okay, can you try doing... control-C or [whatever control is on an apple thingy]-C in the shell, and just running [16:41:11] tail -f /var/log/system.log [16:41:15] And refreshing again? [16:42:51] ok: I get this: [16:42:53] Oct 28 17:42:22 sherah-smith-2 GoogleSoftwareUpdateAgent[47164]: CGSConnectionByID: 0 is not a valid connection ID. [16:42:53] Oct 28 17:42:22 sherah-smith-2 GoogleSoftwareUpdateAgent[47164]: CGSGetOnScreenWindowCount: Invalid connection [16:42:53] Oct 28 17:42:27 sherah-smith-2 Terminal[270]: CGSCopyDisplayUUID: Invalid display 0x41dc9d00 [16:43:01] * K4-713 raises eyebrow [16:43:07] haha srsly [16:43:22] that's all it was Katie! invalid display ox41q94230948234 [16:43:28] problem solved. [16:43:36] Obviously! [16:43:39] heheh [16:43:40] * K4-713 smacks forehead [16:44:04] Hur. Maybe we should do this when I can hover by your monitor and frown helpfully into the back of your head. [16:44:27] ha. yeah. [16:44:54] the only thing is, I think they're wanting it for WorldPay tomorrow. [16:45:01] So, what I normally do (and what happens on the payments cluster) is that we have syslog route messages to different dedicated files based on payment gateway and/or log message severity. [16:45:08] I have *no idea* how that works on osx. [16:45:20] And, yes. I know they're excited about WP. [16:45:25] that makes a lot of sense. [16:45:28] If you've commited these things, I can run them locally. [16:45:38] I've got all my local log buckets sorted out. [16:45:40] yes, they're in gerrit, I think the patches are in that email threat [16:45:42] thread [16:45:46] haha, threat [16:45:54] I did NOT send you an email threat [16:46:08] ...why not? :p [16:46:24] "I WILL PUT A LIVE BADGER IN YOUR FILING CABINET." [16:46:26] I haven't lost sleep yet [16:46:32] haha [16:46:47] I love creative ways to badger people [16:47:00] "HOW DO YOU LIKE ME NOW." [16:47:21] "Love, Katie" [16:47:38] heheh [16:49:20] just a head's up, I am out of here in a few, last dinner plans in Berlin wooo [16:49:32] Fast month, man. [16:49:38] Are you over it yet? [16:49:39] I know… [16:49:58] over Berlin? not at all. I mean, I'm not sad to leave, but I'm not dying to get out of here [16:50:07] it became home [16:50:25] so it feels less whirlwind-romancey and more just cute-homey [16:50:37] exactly what I wanted. [16:50:42] Usually when I travel, I'm like "Woooo! Yay! Awesome! This is Greeeeeat! I NEED TO GO HOME RIGHT NOW." [16:50:52] haha [16:51:05] it will be good to be home though. I kind of miss the office. I like variety. [16:51:16] that's why I'll always be a nomad. [16:51:31] are you dressing up on Halloween? [16:51:50] If I am, it'll be an old costume. But probably one nobody here has seen. [16:51:55] I have not had time this year. [16:52:18] If I'd had time, I'd have done this: [16:52:18] https://www.google.com/search?q=rimmer+holovirus&biw=1794&bih=900&source=lnms&tbm=isch&sa=X&ei=tMlPVNO_EMK4ogST7oEw&ved=0CAgQ_AUoAw&dpr=0.8 [16:52:53] It is stupidly difficult to find a red check gingham dress and matching bonnet in October. [16:53:21] haha hahahha [16:54:22] Okay, so... looking at these patches. Are there just the two? [16:55:02] Gerrit is mad ad your whitespace. [16:55:06] *at [16:57:04] ugh [16:57:26] yes just two, one for DonationInterface and one for the form settings [16:57:57] This is which card? The only one in PCR is a different commit. [16:58:35] Also: Huh, this is new. [16:59:26] All your new ffnames are aliases for the same form, and are enabled for all the same things? [16:59:46] #2083, which I'm still working on [17:00:07] well, when you pass in the ffname it triggers the right partial [17:00:17] as a parameter instead of country [17:00:27] since it's hardcoded [17:00:55] AndyRussG: urrgh, sorry I lied, another meeting happening. [17:01:06] I can chat in an hour if y'd like though! [17:01:15] awight|virtumeet: sure thanks! :) [17:01:36] pizzzacat: Yeah, I figured it must be happening there. Okay. My only comment is that you can define the base and just set the other ffnames to be the same array as the first one, but I'm not going to block on that. [17:01:51] I'll try to put more notes in the GDoc meanwhile, hopefully they'll be more coherent than when I say stuff live [17:02:02]