[06:51] jam, ping [06:51] hi fwereade [06:52] jam, I need to talk with you about constraints at some point [06:52] jam, I'm probably still a bit too tired to do so usefully today [06:53] fwereade: np. You're in EU time now, right? [06:53] jam, but I would be most grateful if you would think briefly about what was painful so we can have a productive chat tomorrow [06:53] jam, yeah [06:55] jam, my rough recollection is that juju-supplied defaults are bad, and the provider should be supplying those itself; this sounds sane, but if there's anything else I'll want to know :) [07:07] rogpeppe, ping [07:29] TheMue, heyhey [07:31] fwereade: Good morning, William. Had a good travel back? [07:31] TheMue, yeah, not bad [07:32] TheMue, frankfurt airport seems to have become nicer somehow [07:32] TheMue, maybe it's just that I was there at 5am on sunday and it seemed like nobody else was :) [07:32] fwereade: Hehe, yeah, a special time. [07:34] fwereade: A few days agon, on Monday, FRA was alost closed due to snow. [07:34] almost [07:35] TheMue, yeah, LHR and JFK were also both snowy, I was pretty lucky with all my travel tbh [07:44] fwereade: pong [07:44] rogpeppe, heyhey [07:44] morning all! [07:52] later guys, might be around off and on but don't count on it, basically calling today a swap day [08:58] re, back from dentist. [09:28] Hmm, somehow my editor dislikes me this morning. Package management for extensions like e.g. for Go fails. [09:49] dimitern: if you see mgz around, let him know I might be a little bit late for our 1:1 [09:49] jam: sure [10:01] mgz: jam asked me to tell you he'll be a bit late for the 1:1 [10:03] dimitern: thanks [10:54] hi mramm, I'm chatting with mgz right now, but when we're done, I'd love to finally catch up with you [10:56] jam: sounds good. [11:25] mramm: poke. I'm on mumble if you are [11:26] I am [11:38] lunchtime [12:02] Greetings [12:04] so, back [12:04] niemeyer: Hiya [12:05] niemeyer: Nice idea, the statistics for baby feeding. ;) [12:07] TheMue: Cheers :) [12:30] core guys, can someone look at my one-line change and confirm it's correct? [12:34] mgz: You've got a +1. [12:38] mgz, hell, sorry about that... we should probably have a trivial test on every package to stop things like this from happening [12:38] morning niemeyer [12:39] (not actually working today) [12:39] fwereade: yeah, I'd like commands to have some "does --help" work thing, but haven't worked out a neat version of the python idiom for that yet [12:40] fwereade: yo [12:40] mgz, good idea [12:40] niemeyer, how's it going? [12:40] fwereade: Good, will start the week having a look at goamz [12:40] niemeyer, excellent [14:25] fwereade: i've got a patch for gotest which complains if this happens [14:26] rogpeppe, sweet [14:27] fwereade: unfortunately i seem to remember it wasn't acceptable [14:28] rogpeppe, haha, ok [14:28] rogpeppe, go build ./... && go test ./... :-/ [14:29] fwereade: isn't it working? [14:29] rogpeppe, I expect it is now, but it wasn't [14:29] fwereade: BTW i've got a workaround script for the fact that you can't do go test -c ./... [14:30] fwereade: How about the still open idea of having a CI server? [14:31] fwereade: ah, that must've been your Conn changes. don't you always run go test ./... before submitting? [14:31] rogpeppe, ofc I do [14:31] rogpeppe, that doesn't help with the packages without tests ;p [14:31] fwereade: ha [14:31] fwereade: hehe [14:32] fwereade: i think go test should at least try to compile the package even if has no tests [14:32] TheMue, +1 on that, not sure when I'm going to get around to it though [14:32] rogpeppe, agreed [14:32] fwereade: i'll raise an issue [14:32] TheMue, wasn't davecheney working on that a while ago? [14:32] rogpeppe, cool [14:33] fwereade: dunno if already working, but imho thinking about it. === slank_away is now known as slank [14:56] * niemeyer => lunch.. biab === slank is now known as slank_away === slank_away is now known as slank [17:30] mramm: Do you know if the consistency call is rolling today? [17:39] niemeyer: I believe that it is not [17:39] mramm: Cool [17:56] just run first tentative benchmark of rpc server, just for sanity check, and i get about 4000 requests a second (or 2000 rpc when round-tripping to mongo). i *think* that should be ok. [17:56] s/rpc/rps/ [17:56] s/2000 rpc/2000 rps/ :-) [17:58] rogpeppe: Sounds nice. [18:00] TheMue: that's not fantastic really - it's an upper bound on how many requests we can expect to serve per second from a single api server instance. [18:00] TheMue: but i think we probably have some possibilities for optimisation [18:11] rogpeppe: I think it's ok as it is your first approach and e'thing runs on only one machine (calling and processing). [18:12] TheMue: yeah, but that means it's actually faster than it will be in real life, i think (we're doing almost no processing). we'll see. [18:14] rogpeppe: I'm optimistic that you may tune it a but. [18:14] bit [18:15] TheMue: we actually have the possibility of using a different protocol for agents (we could maybe use gob, for example) [18:16] TheMue: and that could be quite a bit more efficient than websockets+json, i think [18:16] TheMue: so we do have leeway [18:18] rogpeppe: gob may be nice as the volume is smaller. but how about clients in different langs then? [18:18] TheMue: all our agents are written in the same language :-) [18:18] rogpeppe: thankfully you don't use xml. :P [18:19] rogpeppe: the api is only inteded between the agents and the server? thought of the html client too. [18:19] "test/plain" [18:19] Interesting content-type to find in a test case.. [18:19] TheMue: we could use a different protocol for agents vs html clients [18:19] niemeyer: ha [18:19] rogpeppe: ah, ic, that's in that case pretty nice. [18:19] I almost don't want to fix it :-) [18:20] niemeyer: lol [18:20] niemeyer: :) [18:33] right, all tests passing, time to call it a night [18:33] g'night all [18:34] niemeyer: happy goamz sprinting! [18:34] rogpeppe: Cheers [18:34] bye roger [18:34] rogpeppe: Have a good time there [18:34] niemeyer: and you === slank is now known as slank_away === slank_away is now known as slank