14:28:53 <karsten> #startmeeting metrics team
14:28:53 <MeetBot> Meeting started Thu Jul 12 14:28:53 2018 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:28:53 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:31:47 <karsten> irl: you here?
14:31:53 * irl 
14:32:01 <karsten> great! ready to start?
14:32:05 <irl> sorry, had a phone call unexpectedly
14:32:10 <irl> ready to start
14:32:16 <karsten> no worries. okay, great!
14:32:22 <karsten> * New lookup service (karsten)
14:32:37 <karsten> I saw your response on that thread.
14:32:53 <karsten> it sounded like you liked the idea.
14:33:22 <irl> yes, i think it's a good idea
14:33:53 * karsten tries to find the deadline..
14:34:20 <karsten> july 31, I think.
14:34:43 <irl> do we write the proposal, or some of it, or is that someone else?
14:35:22 <karsten> that's a fine question. but even if someone else does we'll have to make sure it describes what we want to do later.
14:35:44 <irl> ok
14:35:57 <karsten> should we start a new pad for this idea and start writing down our ideas?
14:36:07 <irl> that sounds like a good idea
14:36:34 <karsten> how's tomorrow afternoon as deadline for this part?
14:36:44 <irl> the requirements?
14:36:57 <karsten> the general idea, the benefits, etc.
14:37:09 <irl> ok, sounds good
14:37:11 <karsten> basically, a more comprehensive reply for isabela.
14:37:25 <karsten> okay, I'll start a pad after this meeting.
14:37:32 <irl> cool
14:37:58 <karsten> * Roadmap updates
14:38:16 <karsten> should we go through the list and update progress for all items?
14:38:17 <irl> i noticed that our roadmap on the wiki was looking outdated
14:38:21 <irl> sounds good
14:38:46 <karsten> yes, that's true. it's a bit outdated. though there was not much to update in june.
14:38:49 <karsten> but let's do it now.
14:40:33 <irl> oh dear. this has not pasted well
14:40:44 <karsten> ah, wait, I have a better plan.
14:41:34 <karsten> better.
14:41:37 <irl> yay (:
14:47:39 <karsten> so, anything missing?
14:48:11 <karsten> any why does trac change the 19. to a 10. ...
14:48:37 <irl> i think it just does an <ol> with a starting number maybe
14:48:49 <irl> i think that is all
14:49:09 <karsten> probably. but there must be a way to work around that. I'll find out.
14:49:51 <irl> blank line between them so it's in a new group?
14:50:09 <karsten> yes, maybe.
14:50:26 <karsten> I'll do that after the meeting. no need for you to wait for me reading the manual.
14:50:53 <karsten> okay, next one?
14:50:56 <irl> yep
14:51:00 <karsten> * Onionoo release planning (karsten)
14:51:11 <karsten> we have a few changes already.
14:51:28 <karsten> should we put out a new release in the next days, that is, early next week?
14:51:40 <karsten> and if so, is there anything that should go in that we didn't write yet?
14:52:08 <irl> #6946 is still in need of review
14:52:29 <irl> nothing else that i'm working on at the moment though
14:52:48 <karsten> yes, that review is on my list.
14:53:10 <karsten> okay. then let's just release what we have.
14:53:16 <irl> sounds good to me
14:53:19 <karsten> (including #6946 once it's in.)
14:53:30 <karsten> early next week?
14:53:45 <irl> depends how quickly you can do the review
14:54:02 <irl> but this works for me
14:54:26 <karsten> I think tomorrow is too optimistic, if I also want to write something for isabela.
14:54:42 <irl> i mean, if you're too busy, then later next week is also ok
14:54:44 <karsten> release includes doing the instance upgrade dance, too.
14:54:51 <karsten> ah, early next week should work for me.
14:54:58 <irl> ok, let's do it then
14:55:04 <karsten> alright!
14:55:13 <karsten> * monorepository for all metrics code (karsten)
14:55:19 <karsten> this is a more long-term idea.
14:55:29 <irl> oh no
14:55:42 <karsten> I noticed that we're making the same changes to multiple repositories.
14:56:00 <karsten> and in practice this means we're only doing them to some, not all of them.
14:56:25 <karsten> okay, then maybe we can find a way to avoid this issue with our multiple repositories.
14:56:37 <irl> well, with the current codebase maybe it's a good idea
14:57:04 <irl> but with the new lookup service, i was starting to think about how we might change the architecture of tor metrics into a series of smaller loosely coupled services
14:57:30 <karsten> yes, but that would still be possible with a monorepo.
14:58:04 <karsten> we would just generate 10 different jar files for the various services.
14:58:15 <irl> ...and they wouldn't all necessarily be java services
14:58:17 <karsten> of have 1 jar file and 10 different configurations to do vastly different things.
14:58:22 <karsten> ah, oh.
14:58:28 <irl> (long long term idea)
14:58:36 <karsten> heh. what would they be written in?
14:59:02 <irl> whatever, as long as there is a good reason for it
14:59:36 <irl> i think at the moment it is not very clear what the best interfaces are to use for metrics services, except maybe for onionoo
14:59:55 <irl> collector is best consumed with metrics-lib but in other languages you have per data format rules on filenames, etc
15:00:07 <irl> there's not really a "native" collector interface that we implement with our own tools
15:00:15 <irl> we've always just adopted the interfaces that other tools had
15:00:52 <irl> i wonder if the new bandwidth list format could be generalised to turn it into a generic interface that we could write parsers for, and then have that as the output format for the new lookup service, for example
15:01:17 <irl> then the lookup service provides a webserver to fetch those files from, similar to dir-spec's protocol
15:01:29 <irl> maybe this is a crazy idea though
15:01:45 <karsten> no, it's worth thinking about how to simplify things.
15:02:14 <irl> having seperate code paths should be the "exception" not the "normal" ideally
15:03:03 <karsten> well, okay.
15:03:04 <irl> but now that i think more about it, this would probably require refactoring, and that might be easiest achieved in a monorepo
15:03:20 <karsten> maybe let that idea sink for a bit.
15:03:29 <karsten> no need to make a decision today.
15:03:42 <irl> i will keep it in mind
15:03:55 <karsten> I just observed that some changes are harder to make in all tools than they should be.
15:04:32 <karsten> and please keep thinking about other long-term changes, like languages, etc. we could discuss this more in mexico.
15:04:45 <irl> ok cool
15:05:10 <karsten> so far the main argument for java has been that there's simply so much code in java that changing that would keep us busy for many months without making actual progress.
15:05:39 <karsten> anyway.
15:05:56 <karsten> did we run out of topics?
15:06:01 <irl> i have no more topics
15:06:24 <karsten> alright! looks like we have action items on the pad and in our inboxes.
15:06:57 <karsten> thanks, and bye!
15:07:01 <irl> bye!
15:07:02 <karsten> #endmeeting