15:00:05 #startmeeting metrics team 15:00:05 Meeting started Thu Jan 19 15:00:05 2017 UTC. The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:05 Useful Commands: #action #agreed #help #info #idea #link #topic. 15:00:12 hi iwakeh, RaBe, luketheduke! 15:00:47 * karsten opens the agenda pad: https://pad.riseup.net/p/3M7VyrTVgjlF 15:01:52 anything we should add there before we start? 15:03:23 alright, let's go. 15:03:29 - Web design (RaBe) 15:03:50 how's that going? I noticed you wrote an atlas patch, 15:03:59 hi 15:04:03 and you started looking into javadocs css. 15:04:06 hi hiro! 15:04:11 hiro: Hi! 15:04:15 Yeah... Just to start, I took a very low priority ticket to see how all of this works, so please let me know if this is how it's done correctly or if you'd prefer me to do it some other way :) https://trac.torproject.org/projects/tor/ticket/12685 15:04:27 good thinking! 15:05:35 I didn't look at the code, but the process (trac comment) looks good to me! 15:05:56 irl: you around? 15:05:57 as you see, i just cloned the rep to github, made a branch for that task there, and linked the commit in the trac ticket 15:06:03 minimal invasive one-liner -) 15:06:22 I wonder if irl would prefer tickets to be reassigned to him for review. 15:06:32 if this works for you, i'll do multiple lines next time :) 15:06:36 hehe 15:06:44 the fewer the better :-) 15:06:52 the tickets was assigned to phw, so i didn't know if i should change that 15:07:00 right, I don't know either. 15:07:31 am i allowed to work on ANY ticket that is set to ANY user, or just IRLs tickets? 15:07:33 what you could do, if we don't hear anything now, is cc him. 15:07:55 ticket owner is not as meaningful here. 15:08:00 please feel free to reassign any atlas ticket that's assigned to me. 15:08:07 great :) 15:08:11 hi phw! :) 15:08:17 hey karsten :) 15:08:27 we could do a batch reassignment if irl is okay with that. 15:08:45 there, enough highlights for now. :) 15:08:54 just let me assign the tickets to myself that i'm working on :) 15:09:09 yes. 15:09:24 and cc irl, if he not on there. 15:09:32 works for me, too. 15:09:37 okay 15:09:47 Also, I had a look into the JavaDocs and updated the stylesheet file a bit with the tor style guide colors and metrics page looks: http://javadocs.cc-ltd.net/descriptor/ 15:10:11 nice! 15:10:15 i can't do anything i want because i'm limited to the css file (html changes would be needed for some other stuff) 15:10:38 however i think this looks more like tor than ever before ^^ 15:10:40 * iwakeh trying to call the page 15:11:12 that's fine. no need to make it perfect, but adapting to colors is a great start. 15:11:26 less is more there, too. 15:11:28 Hi guys, apologies, I'm a bit late :) 15:11:33 Yes, that looks more like Metrics ci. 15:12:02 i could include an onion logo, but that would have to be on another server, outside the javadoc itself, and wouzldn't work offline 15:12:21 we could include the logo in the source. 15:12:28 rehi 15:12:47 hi nickm! 15:12:53 alecmuffett: oh! I was not arguing against single onions! 15:12:53 iwakeh: does this work with automated generated javadocs? 15:13:24 alecmuffett: I was arguing that, _if_ we were dogmatically opposed to user options, we wouldn't have them. 15:13:30 RaBe: didn't try it recently, but these things can be specified when generating javadoc. 15:13:51 alecmuffett: (but we aren't, and we do) 15:14:29 iwakeh: do you have a link, maybe? 15:14:35 iwakeh: ok, lets try that and start with a logo. also i would not mind if we could include the correct font we use for the metrics page :D 15:14:39 RaBe: we could also simply add the logo file to the javadoc output. 15:14:58 * iwakeh looking for a link ... 15:16:22 https://docs.oracle.com/javase/8/docs/technotes/guides/javadoc/index.html 15:17:15 hhm, that not it. 15:17:17 hmm, they're talking about "any text file output" 15:17:46 we'll find out later. i can prepare an optional logo :) 15:17:55 yes, sounds good. 15:18:02 maybe, http://www.oracle.com/technetwork/java/javase/documentation/index-137483.html#usingHead 15:18:10 helps a little. 15:18:11 I'd say don't sink too much time into that. if it's easy, feel free to play around with it. 15:18:18 right. 15:18:34 okay! 15:19:00 okay, anything else we can discuss now to make your life easier in the next week? :) 15:19:05 RaBe: we can put a logo file in anywhere in the output directory. 15:19:21 iwakeh: ok great! 15:20:02 RaBe: actually, any type of file. 15:20:24 karsten: i think i ll have a look in some of the tickets until next week, and then we start the other projects? or do you want to discuss them now? 15:20:48 should be fine next week. 15:20:57 if that's okay for you. 15:21:42 yes, absolutely! 15:21:46 cool! 15:21:52 good stuff :) 15:22:00 let's move on to some performance measurements. 15:22:05 - OnionPerf (hiro) 15:22:11 yes 15:22:22 so many options there... :) 15:22:29 :) so I am setting things up on otf cloud and see if I run into any issues 15:22:38 okay, great! 15:22:55 I got feedback from Rob on the readme last week 15:23:23 but I'd like to wait before merging I want to see some files and measurements after a few days that we run onionperf 15:23:31 yes, that makes sense. 15:23:50 one thing to consider: 15:24:02 at some point we'll need to collect measurements somehow. 15:24:28 I lost track what support onionperf has for that, or whether we'll have to write something ourselves. 15:24:41 so onionperf generates xml logs 15:24:44 maybe that's something you can start thinking about while you're setting up these instances. 15:24:52 and then you can generate pdfs 15:24:59 so that's not ideal for a webapp 15:25:04 * irl was interviewing candidates to take away some of his work, sorry it overran, reading backlog 15:25:12 heh, irl. hi! 15:25:17 we'd need logs or files with measurements. 15:25:35 ideally we could parse the logs and have a small web interface 15:25:39 the pdfs won't be used in metrics. 15:25:42 hiro: so, onionperf also has a feature to export measurements in the torperf format. 15:25:56 yes that too 15:26:23 that's what we need. 15:26:35 at least that's easiest to integrate for now. 15:26:59 though we'll still have to write code to collect those files. 15:27:13 because currently collector expects .data and .extradata files and merges them itself. 15:27:16 that's ok for me. what we could do is have routine that exports the logs every fixed amount of time 15:27:16 it's a mess. 15:27:42 right, something like that. 15:27:43 would be good to know the structure of such files and what options there are to tweak the file output. 15:27:58 agreed. 15:28:02 we could also add some code to onionperf to output what we want 15:28:13 so we do not have to parse the logs 15:28:23 rather keep it in collector. 15:28:34 but, are there only logs? or 15:28:47 real measurement output files? 15:29:42 Quoting: hiro: so, onionperf also has a feature to export measurements in the torperf format. 15:30:15 how about we move this discussion to next week? I just thought I'll bring up the topic here so that we start making plans. 15:30:22 ok 15:30:24 fine. 15:30:39 * irl is caught up on backlog 15:30:47 great! hoping for the otf cloud thing to work! 15:30:59 yes! 15:31:37 irl: if you look at that low priority ticket, is this working for you? :) 15:31:52 ok will let you know how that goes 15:32:05 hiro: great! thank you. :) 15:32:09 RaBe: can you set the owner to me when you set it to needs_review? this way it shows in my todo list and i get a mail 15:32:14 should we have a ticket 15:32:23 for the onionperf discussion? 15:32:24 irl: no problem! 15:32:32 iwakeh: ticket sounds great! 15:32:36 RaBe: also please reassign any tickets you are working on or planning to work on soon to yourself 15:32:57 RaBe: my local task list is built by scraping the trac, so this gives me a clearer view of the ones i need to look at 15:33:17 ok I can create that 15:33:27 Thanks! 15:33:41 * karsten quickly adds a few things to irl's todo list....... ;) 15:34:03 irl: okay! 15:34:22 okay, sounds like we have plans for those two agenda items. moving on? 15:34:44 sure. 15:34:52 - Tech reports (iwakeh) 15:35:01 curious about those. :) 15:35:08 writing and trying to stay away from code. 15:35:16 heh 15:35:35 Digging through Metrics. 15:35:59 that's why those tickets come up here and there. 15:36:03 I think 15:36:12 yes, I figured that must be where they come from. 15:36:18 hehe 15:36:44 I'm aiming at having the basis for the first two reports and 15:37:08 so the third will have something to refer to. 15:37:31 sounds good. 15:37:46 Estimating is more difficult, but I'd like to send some drafts soon. 15:38:01 estimating network totals? 15:38:12 ? 15:38:20 what do you mean by estimating? 15:38:52 estimating the time to completion, the content referring 15:38:55 ah! 15:39:08 to estimates is rather easy compared to the time ;-) 15:39:15 I think it's more important that these reports are useful than done really soon. 15:39:28 good to hear! 15:39:29 take your time, I'd say. 15:39:46 of course, feel free to share drafts if feedback would be useful. 15:40:07 I will, keep asking questions and creating tickets. 15:40:12 yes! 15:40:31 alright, moving on? 15:40:40 yep. 15:40:45 - Onionite launch (lukechilds) 15:41:17 lukechilds: would you want to write a sentence for https://metrics.torproject.org/operation.html ? 15:41:51 I was also thinking whether we should add screenshots/thumbnails there. the list is rather long now, and maybe something visual would help. 15:41:58 i.e., what is Onionite in three words. 15:42:01 this list in particular: https://metrics.torproject.org/operation.html#status 15:42:48 I can add screenshots/thumbnails. but I need that short description for onionite. 15:42:59 yeah sure 15:43:08 do you mean literally one sentence? 15:43:15 yes. 15:43:15 like the others. 15:43:24 starting with Onionite. :) 15:43:37 cool 15:43:51 yeah you can just take the discription straight from github 15:43:53 PWA to view information on the individual nodes that make up the Tor network. 15:44:07 or "Onionoo is..." 15:44:23 to satisfy your starting with Onionite requirement ;) 15:44:34 somewhat related, should we remove this list here? https://onionoo.torproject.org/ 15:44:51 yes. 15:45:01 rather refer back to metrics.tpo 15:45:06 what's a PWA... 15:45:14 I think the API clients list may be useful to keep there 15:45:22 Progressive Web App 15:45:39 what's a progressive web app. okay, I can search for that. :) 15:45:40 Good mobile support and works offline 15:45:57 okay, I'll take that. 15:45:57 You can install it to your phone and use it like a native app 15:46:35 regarding the list of API clients, that's why we have the lists on metrics.tp.o. 15:46:38 also, https://metrics.torproject.org/development.html 15:46:55 we moved everything from the onionoo page there. or at least we tried. 15:47:04 yes, Onionoo should just link here. 15:47:13 Fair enough, I suppose it makes more sense to have everything in one place 15:47:23 less maintenance. 15:47:28 yes. 15:47:37 maybe you want to take a look if anything is missing? 15:47:59 what about thumbnails/screenshots? should I give that a try? 15:48:29 I got disconnected... 15:48:43 I can do them if you want 15:49:08 do you jsut want one of the home page? 15:49:09 ah, sure, if you want to do that, please do it. 15:49:11 hiro: you didn't miss anything. 15:49:15 I don't know. 15:49:41 I already have this 15:49:41 something that's easy to do for the next 5 websites. 15:49:43 https://camo.githubusercontent.com/330a88159d74ee6ca84823cf8e1ba3555e236e0c/68747470733a2f2f692e696d6775722e636f6d2f565961564b686d2e706e67 15:50:00 ok, maybe not that then :) 15:50:15 or maybe we can ask RaBe to think about that for a few minutes. 15:50:59 meeting time is running ... ;-) 15:51:06 it is. :) 15:51:12 moving on: 15:51:15 sorry, i'll have to read the backlog :) 15:51:16 - Tor Browser downloads and updates (karsten) 15:51:20 thanks, RaBe! 15:51:50 so, there's a ticket for this.. 15:52:16 https://trac.torproject.org/projects/tor/ticket/21236 15:52:38 we're mostly talking about code things there, that won't affect the graph. 15:52:55 I wonder how we should discuss the graph. 15:53:04 or whether there's much to discuss anyway. 15:53:17 well, admittedly I didn't look at the graph yet. 15:53:26 maybe we can involve applications team folks in that discussion. 15:53:40 would you want to take a look before we ask more broadly? 15:53:54 yes, they have ideas what they want to see there, I assume. 15:54:06 probably. :) 15:54:21 we're quite limited with parameters on the current metrics website. 15:54:22 Just ask there surely be changes 15:54:45 but there might be other changes to the wording, for example. 15:54:59 or maybe we picked the wrong requests for something. 15:55:03 request strings, that is. 15:55:14 okay, should I wait for you to look at the graph first? 15:55:30 well, I look at it after this meeting 15:55:35 and add a comment. 15:55:36 perfect! 15:55:40 thanks. :) 15:55:45 * irl needs to vanish again for the next meeting, RaBe: will review anything set to needs_review after i'm home and have had dinner, and will merge if it's good (: thanks for your work 15:55:46 sure. 15:55:57 thanks, irl! 15:56:06 bye, bye, irl. 15:56:18 last item on the agenda: - Planning for Berlin meeting in Jan/Feb (karsten) 15:56:27 irl: there's just that one ticket, but there's more to come :) bye! 15:56:46 assuming we'll have such a meeting, what would we discuss there? 15:56:48 or work on? 15:57:11 I guess we should start making such a list. 15:57:23 on the pad? 15:57:29 sure, on the pad would work. 15:57:36 or e-mail? 15:57:42 or that! 15:57:50 who'll attend? 15:58:21 I assume you, hiro, and I. maybe juris, but mostly to say hi. 15:58:31 fine. 15:58:38 timeframe? 15:59:06 I probably have the shortest time getting there :-) 15:59:14 I think early feb is more realistic than late jan. 15:59:21 agreed. 15:59:26 :) also works better for me 15:59:28 later in feb might be less useful, because it's closer to the amsterdam meeting. 15:59:35 true. 16:00:02 okay. email? 16:00:16 yep. fine with me. 16:00:21 yep email is easier 16:00:26 great! 16:00:32 does that mean we almost finished on time?! 16:00:39 otherwise we have to remember to keep looking at the pad for updates 16:00:42 seems like it 16:00:45 right. 16:01:00 awesome. thanks, everybody! bye! :) 16:01:03 #endmeeting