15:00:31 <karsten> #startmeeting metrics team
15:00:31 <MeetBot> Meeting started Thu Feb 16 15:00:31 2017 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:31 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:00:40 <karsten> - Atlas usability project and blog post (RaBe)
15:00:55 <karsten> RaBe: I went ahead and put that on the agenda.
15:01:02 <karsten> hope that was okay.
15:01:12 <karsten> want to give a quick update how this is going?
15:01:30 <karsten> irl: same question to you, I guess.
15:01:51 <RaBe> oh, unfortunately nothing new in the past two weeks - irl was in spain, and I was working on CollecTor and other stuff :)
15:01:55 <irl> 7 tickets ready for review, 1 ready for merge
15:02:07 <karsten> neat!
15:02:15 <irl> i'll be catching up on those reviews this weekend
15:02:35 <irl> will start on drafting a blog post, ready for the end of the month
15:02:36 <karsten> the plan is still to have a blog post by end of the month?
15:02:39 <irl> yep
15:02:39 <karsten> awesome!
15:03:03 <karsten> feb 27 is when we're going to launch onionoo protocol version 4.0.
15:03:20 <karsten> we could aim for the 27th or 28th for the blog post.
15:03:26 <irl> this sounds good
15:03:34 <karsten> great!
15:04:00 <karsten> okay, if there's nothing to discuss today, shall we move on?
15:04:14 <karsten> (nothing related to atlas, that is...)
15:04:42 <irl> sounds good, RaBe: we can continue as before now that i'm back
15:04:43 <karsten> please bear with me today, antibiotics are running this body...
15:04:51 <RaBe> irl: sure!
15:05:18 <karsten> okay, great!
15:05:35 <karsten> moving on to the next topic:
15:05:37 <karsten> - CollecTor website (RaBe)
15:05:43 <karsten> I just commented on the ticket.
15:05:53 * iwakeh just commented on karsten's comment.
15:05:57 <karsten> but maybe we should look at the prototypes together.
15:06:05 <iwakeh> '21381
15:06:09 <iwakeh> #21381
15:06:10 <RaBe> i just read ticket #21381... I'm not sure if i understand the suggestion with the "big table with buttons"
15:06:42 <karsten> you mean this one? https://collector.torproject.org/#available-descriptors
15:07:05 <karsten> if you click on one of the "format" buttons, you'll navigate to the data formats section further down below.
15:07:24 <RaBe> oh, i never figured that one out :)
15:07:36 <karsten> hehe. can we make it more obvious somehow?
15:08:02 <iwakeh> whoever goes there for the contents will notice, I think.
15:08:27 <RaBe> i think that was just me... i agree with iwakeh
15:08:28 <iwakeh> But, that's hard to notice when just looking at the layout.
15:08:55 <karsten> okay.
15:09:02 <RaBe> we could add an ">" to clearify they're buttons?
15:09:21 <karsten> sure, feel free to try something like that.
15:09:55 <RaBe> so if i understand all comments correctly, we won't have ANY navigation left at that page?
15:10:07 <RaBe> (beside that table...)
15:10:09 <karsten> could be. :)
15:10:17 <iwakeh> the table is the navigation.
15:10:33 <iwakeh> and browser search.
15:10:46 <karsten> maybe we could turn the table more into a table of contents.
15:10:52 <karsten> not sure how to do that.
15:10:58 <iwakeh> well,
15:11:14 <iwakeh> the descriptor types could become links.
15:11:21 <karsten> yes, or that.
15:11:28 <iwakeh> to the format.
15:11:41 <RaBe> i like that idea
15:12:02 <karsten> another thing we could do is do something with cell background colors, indicating that a line is a relay descriptor, a bridge descriptor, etc.
15:12:22 <karsten> not sure how that would work. most entries are for relay descriptors.
15:12:33 <hiro> wouldn't that be too many colors?
15:12:35 <karsten> or maybe something with thicker lines between groups.
15:12:46 <karsten> yes, I guess I mean two alternating colors.
15:12:49 <RaBe> what about labels, like we use on the news?
15:13:02 <RaBe> or "badges"
15:13:22 <iwakeh> or group headers, "relay descriptors"
15:13:32 <karsten> I think rather group headers than badges.
15:13:34 <iwakeh> "bridge descriptors" etc.?
15:13:39 <karsten> badges might be confused with buttons.
15:13:44 <karsten> well, assuming we keep those.
15:14:08 <karsten> RaBe: want to take this input and try some things?
15:14:08 <iwakeh> the format buttons will disapear.
15:14:20 <karsten> yes, that one will disappear. I meant the recent and archive buttons.
15:14:41 <iwakeh> maybe, have the column header "access" for the buttons?
15:14:51 <RaBe> i agree. i'll split the table into subsections somehow and comment on the ticket :)
15:15:32 <karsten> okay. quite a few options there.
15:15:56 <karsten> shall we move on?
15:16:01 <iwakeh> yep.
15:16:15 <karsten> - Tech reports (iwakeh)
15:16:28 <karsten> yay, I saw a draft today. :)
15:16:45 <iwakeh> yes, the second. I'm assembling some ideas and questions, which I'll mail later today.
15:16:53 <iwakeh> with
15:17:05 <iwakeh> ideas about how to proceed more speedily.
15:17:13 <karsten> sounds good!
15:17:20 <karsten> again, it's quite the topic.
15:17:35 <iwakeh> big
15:18:00 <iwakeh> any other references I should take note of?
15:18:23 <karsten> fine question. I mention them as they come to mind.
15:18:31 <iwakeh> ;-)
15:18:34 <karsten> under the assumption that more input must be better.
15:18:44 <karsten> and that you filter input appropriately.
15:18:46 <karsten> ;)
15:19:07 <iwakeh> yes, that filtering will
15:19:14 <iwakeh> work some way :-)
15:19:19 <karsten> heh
15:19:33 <karsten> alright, looking forward to that mail later today.
15:19:49 <karsten> moving on?
15:19:51 <iwakeh> I'll try to keep it short
15:19:56 <iwakeh> yes.
15:20:01 <karsten> - OnionPerf (hiro)
15:20:05 <hiro> sure
15:20:30 <hiro> so last week I was finishing running some quick test to see if our onionperf deployment was prducing the same results as torperf
15:20:41 <hiro> it looks like it is.
15:20:48 <karsten> neat!
15:20:58 <hiro> I basically did the parsing myself and used a python notebook to do a couple of plots
15:21:07 <karsten> can you share those?
15:21:15 <hiro> sure
15:21:28 <hiro> I am going to send an email after this meeting
15:21:32 <karsten> okay, great!
15:21:39 <iwakeh> or, attach to that ticket?
15:21:49 <karsten> how many instances are you running?
15:21:54 <iwakeh> otherwise it get's lost in inbox too easily.
15:22:05 <hiro> ok I will attach to the ticket better
15:22:13 <karsten> sounds good.
15:22:20 <iwakeh> thanks, that's easier to comment.
15:22:23 <hiro> I am running only one instance at the moment. So moving forward there are two action items
15:22:49 <hiro> 1. Running also on tpo infrastructure. We have the VM I will set it up after this too
15:23:22 <hiro> 2. What do we do with this data? Shall I work to have it included in metrics-web? Shall I maintain a notebook that people can quickly play with?
15:23:49 <karsten> re 2: I guess we should extend collector to fetch those files and archive them.
15:24:14 <karsten> right now it only knows how to download the older .data and .extradata files and merge them into .tpf files.
15:24:28 <karsten> but downloading .tpf files from remote servers shouldn't be that hard.
15:24:37 <karsten> iwakeh: what do you think?
15:24:42 <hiro> karsten yes. I guess that should be our goal.
15:24:44 <iwakeh> that'll be easy.
15:24:51 <karsten> happy to avoid httpurlconnection. ;)
15:24:55 <iwakeh> hehe
15:25:12 <karsten> hiro: the files are publicly available via http(s)?
15:25:16 <hiro> yep
15:25:34 <karsten> as one big file that gets bigger and bigger, or as one file per day?
15:25:43 <hiro> one file per day
15:25:53 <iwakeh> we need a description of how these files will accumulate and where,
15:25:55 <karsten> cool! can you paste a url to the ticket, too?
15:26:01 <hiro> sure
15:26:09 <iwakeh> i.e., what path (if at all) under which url.
15:26:12 <hiro> each measurement has a different file
15:26:18 <hiro> sure I will add it to the ticket
15:26:55 <karsten> okay, sounds doable then.
15:27:16 <karsten> re 1: are you planning to run a second instance in a different location of the cloud?
15:27:29 <hiro> we can do that too
15:27:37 <karsten> that would mean we'd have 3.
15:28:00 <karsten> by the way, we'll need to come up with names for these instances.
15:28:00 <hiro> sure so I'll add that to my list
15:28:11 <karsten> could be the physical location.
15:28:22 <hiro> at the moment I have onionperf-one and was planning to just increase the numbering
15:28:26 <karsten> or really anything. the current ones are torperf, siv, and moria. hmm.
15:28:46 <karsten> well, numbers work, too. :)
15:28:48 <karsten> anyway.
15:28:50 <hiro> also I should share access to this instances ;)
15:29:16 <karsten> either that, or be around when they break. which works for me, too.
15:29:18 <karsten> :)
15:29:32 <karsten> but yes, having a second operator would make sense in the longer term.
15:29:34 <hiro> ok :) I can do both
15:29:45 <karsten> similar to all the other services.
15:29:50 <iwakeh> http(s) access is sufficient :-)
15:30:05 <karsten> alright, moving on?
15:30:41 <karsten> - metrics-lib 1.6.0 release (karsten)
15:30:49 <karsten> I'd like to put out a release.
15:30:55 <iwakeh> fine.
15:31:01 <karsten> https://gitweb.torproject.org/metrics-lib.git/tree/CHANGELOG.md already has quite a few changes.
15:31:15 <karsten> and there's #20323.
15:31:28 <karsten> well, which could as well go into 1.7.0.
15:31:30 * iwakeh has review planned.
15:31:37 <karsten> okay!
15:31:42 <karsten> how about release tomorrow?
15:31:48 <iwakeh> what about #20430?
15:31:52 <karsten> so,
15:32:03 <karsten> I started looking into that and found that it's related to #16225.
15:32:21 <karsten> and then I started thinking about a better DescriptorSource interface.
15:32:39 <karsten> which I added to #16225.
15:32:42 <iwakeh> Well, ticket description and content differ quite a bit?
15:33:04 <karsten> well, logging is just one way of handling an error.
15:33:21 <karsten> the question is whether it's the right way to handle the error at all.
15:33:26 <iwakeh> I'm referring to the interface changes.
15:33:48 <karsten> you mean the builder pattern?
15:33:51 <iwakeh> but not too quick, first 1.6.0
15:34:03 <karsten> right. so, my plan was to discuss all that for 1.7.0
15:34:17 <iwakeh> ?
15:34:24 <iwakeh> oh, no 1.6.0?
15:34:27 <karsten> logging and unifying error handling.
15:34:41 <karsten> so, my plan was to put out 1.6.0 with what we have.
15:34:56 <karsten> and move the logging improvements to 1.7.0 together with better error handling.
15:35:00 <karsten> including those interface changes.
15:35:08 <iwakeh> ok, depending on the review of 20323, if that'll be 1.6.0 or 1.7.0.
15:35:25 <karsten> either would work, I guess.
15:35:29 <iwakeh> ok
15:35:50 <karsten> but I think that with a better interface, we'll have an easier time writing tutorials and that blog post.
15:36:02 <karsten> the current interfaces are a bit clumsy.
15:36:07 <iwakeh> there are some other tickets
15:36:16 <iwakeh> with similar intentions.
15:36:19 <karsten> true.
15:36:25 <iwakeh> I'll look for them after meeting.
15:36:30 <karsten> yes, please do.
15:36:52 <karsten> okay, sounds like we have a plan for 1.6.0.
15:36:53 <iwakeh> and take a look at 16225
15:37:01 <karsten> yes, please.
15:37:04 <iwakeh> and the planning of 1.7.0
15:37:07 <iwakeh> :-)
15:37:13 <karsten> so many plans!
15:37:25 <iwakeh> well, we get them done, too :-)
15:37:25 <karsten> okay, moving on? :)
15:37:31 <karsten> yes!
15:37:34 <iwakeh> fine.
15:37:39 <karsten> - Berlin meeting (When?)
15:37:49 <karsten> speaking of plans that did not work out..
15:38:05 <karsten> - Agenda planning for AMS (or schedule an online meeting?)
15:38:05 <iwakeh> oh, that's life
15:38:14 <karsten> yep.
15:38:26 <hiro> ams could be a good time
15:38:35 <karsten> I think that's a good suggestion, to make some plans for AMS.
15:38:42 <iwakeh> ams := amsterdam
15:38:56 <karsten> we'll have a team day, I think.
15:39:08 <karsten> where we can schedule a session or two.
15:39:14 <karsten> it's in my inbox somewhere.
15:39:24 <iwakeh> ok, good place :-)
15:39:29 <karsten> hehe
15:40:11 <karsten> okay, how about I dig out what I have in my inbox related to AMS planning and our earlier TXL plans?
15:40:13 <iwakeh> should we have a meeting (irc) next week about that?
15:40:21 <iwakeh> BER
15:40:27 <karsten> is that a thing yet?
15:40:39 <hiro> what's TXL?
15:40:40 <iwakeh> no, BER is any airport in Berlin
15:40:41 <karsten> a separate meeting?
15:40:45 <karsten> ah!
15:40:49 <karsten> TXL = tegel airport.
15:40:51 <hiro> ahh
15:40:54 <iwakeh> txl:= tegel (berlin)
15:41:14 <karsten> or should we add an agenda item for next week's meeting?
15:41:19 <iwakeh> sxl:= schönefeld
15:41:33 <hiro> is there anything urgent that it's better we start discussing before amsterdam?
15:41:33 <iwakeh> yes.
15:41:59 <iwakeh> nothing, which we are not yet discussing, afaik?
15:42:13 <karsten> I can't think of anything in particular, no.
15:42:23 <karsten> we're already preparing for AMS in a way.
15:42:43 <karsten> did you already book flights?
15:42:48 <hiro> ah ok :)
15:42:58 <iwakeh> not yet, thinking about driving maybe.
15:43:02 <karsten> and put your arrival and departure dates on the wiki?
15:43:04 <hiro> my question was related to the question: shall we have a separated  irc meeting
15:43:12 <hiro> yes! I have to do that
15:43:40 <karsten> regarding the separate irc meeting, I think we should try putting it on the agenda for next week and prepare a bit via email.
15:43:51 <iwakeh> sounds fine.
15:44:04 <hiro> yep sounds fine to me too
15:44:09 <karsten> okay!
15:44:28 <karsten> does that mean we're done early today?
15:44:35 <hiro> yay!
15:44:51 <iwakeh> well, on time as we intended this for 45min once upon a time :-)
15:44:58 <karsten> hah, true.
15:45:20 <karsten> alright. thanks for an efficient meeting, everyone!
15:45:33 <iwakeh> thanks! bye, bye.
15:45:38 <RaBe> bye!
15:45:39 <karsten> talk to you next week as usual. bye, bye!
15:45:41 <hiro> thanks! talk later by email/ticket update
15:45:45 <karsten> yep!
15:45:48 <karsten> #endmeeting