14:28:47 #startmeeting metrics team
14:28:47 Meeting started Mon Aug 27 14:28:47 2018 UTC. The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:28:47 Useful Commands: #action #agreed #help #info #idea #link #topic.
14:28:52 irl: around?
14:28:58 yes
14:29:10 cool! let me put some topics on the pad..
14:30:49 do you have anything else?
14:31:02 ok, i do not have any topics to add unless you wanted to talk about anything in the big review
14:31:43 no need to talk more about the big review, unless there's more that needs discussion.
14:31:52 great that you did the review this fast!
14:32:04 it was just the
thing but it's not a blocker
14:32:11 other than that, I think that we'll have to do some serious testing when putting out the next releases.
14:32:47 all in all I think the code is in better shape now.
14:32:50 yep
14:33:14 alright, then let's go through my topics.
14:33:26 * metrics-team post with bug report (karsten)
14:33:33 there was a bug report on the list.
14:33:39 about searching for host names?
14:33:46 yes.
14:33:51 it's the same issue as #27187
14:33:52 I only looked briefly. it looked real.
14:33:56 it is real
14:33:57 ah!
14:34:07 okay, I did not make that connection.
14:34:18 the details documents are all correct now and have all the details, the summary document is not updated and so the indexer is not updated
14:34:33 okay. do you mind replying on the list, pointing to the ticket?
14:34:47 ok, i will do that once i've added the info to the ticket also
14:34:54 perfect!
14:35:07 * instant vs localdatetime vs zoneddatetime (karsten)
14:35:16 so, last week I briefly looked into java 8 changes.
14:35:26 one thing I did not touch is related to dates/times.
14:35:44 the blocker was that I wasn't sure whether zoneddatetime is even the right type.
14:35:55 previously, iwakeh suggested localdatetime.
14:36:07 but I'm also not sure whether we'd be better off using instant. heh.
14:36:18 localdatetime is definitely wrong because it's ambiguous
14:37:01 i guess there are hard questions we need to answer
14:37:03 yes, that sounds plausible. though I briefly thought about documenting that we're always using UTC as timezone.
14:37:08 does tor believe in leap seconds?
14:37:33 I don't know.
14:38:00 instant seems to use the "java time" which is defined differently to UTC
14:38:11 oh?
14:38:27 I thought it just uses UTC and nothing else.
14:38:35 but, I could be wrong.
14:38:41 in UTC there are days with leap seconds where actually seconds are shorter in time than other days
14:39:06 ntp will do this for you but some people prefer to just have clocks jump a second instead
14:39:29 okay, I was thinking about how to approach this.
14:39:37 (I did not intend to resolve this today.)
14:39:58 would it help to go through the codebases and find all different ways how we're using date/time data?
14:40:08 i think a tor-spec laywer needs to see what tor believes and does not believe about time
14:40:52 so, okay, assuming we do that,
14:40:56 but yes, a list of how we currently use time is helpful too
14:41:07 we'll still have to decide what to do with descriptors/data that don't come from tor.
14:41:21 we could argue that those follow tor, though.
14:41:47 so, how about I make such a list, and then we discuss next time whether/how we want to involve network team folks?
14:42:11 there should be mappings from one time into another, but ideally we would convert to one canonical time once it enters tor metrics
14:42:24 that sounds like a good plan
14:42:33 okay.
14:43:35 alright,
14:43:37 * exonerator release with #27266 (karsten)
14:43:57 once this ticket is reviewed (no big rush), we'll have to put out a release.
14:44:12 this will also be the first release to ship with all those little code changes.
14:44:21 when is a good time to make this release?
14:44:59 tomorrow is probably the only full day i have this week, wednesday will be rushed and then thursday i'm on the train to emfcamp
14:45:11 i can review the change this afternoon and we can release tomorrow?
14:45:18 sure, if that works for you, great!
14:45:21 cool
14:45:49 * planned exonerator database change (karsten)
14:45:54 this is mostly a heads-up for now.
14:46:05 I'm still working on the big exonerator database schema update.
14:46:14 for performance?
14:46:24 smaller database and performance, yes.
14:46:31 right ok
14:46:44 the migration script still contains a bug that I just found.
14:47:06 I also sent a list of questions to Sebastian who wrote most of the original changes.
14:47:30 after that I'll do another test run here, which can easily take days.
14:47:56 one question is: would you want to do a review while the test is running or afterwards?
14:48:06 it's not a tiny change.
14:48:33 or would you prefer Sebastian to do the review, assuming that he has time to do it?
14:48:46 i guess as soon as it's something that can be reviewed you can set it to needs_review and then i can pick it up
14:49:01 sounds good to me.
14:49:07 i'm not familiar at all with exonerator's codebase or schema so it would be good if Sebastian can take at least a quick look
14:49:16 okay.
14:50:23 * planned onionoo 7.0 release next week (karsten)
14:50:36 we said after Sep 3.
14:50:46 which can easily be a few days later.
14:50:54 in any case, that's probably next week.
14:51:25 i will still be in a field on sep 3
14:51:43 no worries. is the second half of next week better?
14:51:55 sep 7 would be my first full day back but i may be catching up still if anything urgent has come up
14:52:10 okay, then let's maybe aim for the week after.
14:52:12 no rush here.
14:52:48 we could do monday 10th sep
14:53:05 okay.
14:53:19 are we going to have a team meeting next thursday, by the way?
14:53:25 or should we move that?
14:53:56 i think i will be on the train back
14:54:03 so probably we should move it
14:54:18 by one day to friday?
14:54:26 sounds good
14:54:40 same time?
14:54:43 yep
14:54:50 great!
14:54:57 last topic: any volunteers around?
14:55:15 and by that I mean: any volunteers who mentioned on a mailing list that they'd like to volunteer?
14:55:40 if not, that's fine. it was very short notice.
14:55:55 regarding python
14:56:02 yep?
14:56:08 there is a ticket about adding support for collector's index.json to stem
14:56:23 if someone wants to do metrics and really wants to also do it in python, that would be a good ticket to take on
14:56:32 i can see it being useful to researchers
14:56:40 good idea.
14:56:49 do you have the number?
14:57:16 #17979
14:57:49 atagar even considers the lack of this module a defect, which is nice (:
14:58:03 heh.
14:58:14 alright. I don't have anything else.
14:58:35 what about you?
14:58:57 nothing more from me
14:59:14 alright!
14:59:38 then let's talk more next week. (and do the release tomorrow, asynchronously.)
14:59:47 thanks, and bye!
14:59:51 bye!
14:59:52 #endmeeting