[04:21] le sign [04:21] sigh [04:21] us-east-1 is 260ms from me [04:21] ap-southeast-1 is 160-180 [04:21] hardly rocking === philballew_ is now known as philballew [08:57] fwereade: mornin' [08:57] heya rogpeppe === mcclurmc_away is now known as mcclurmc [09:55] morning TheMue [09:55] heya fwereade [10:30] moin. [10:32] Moin Aram [10:41] Aram, TheMue, mramm: hiya [10:41] rogpeppe: hiya [10:42] hi rogpeppe [10:42] rogpeppe: early morning for me (5:40 here) [10:42] mramm: where are you now? lost between copenhagen and you home town? [10:42] mramm: can't sleep? or inconvenient meeting time? [10:42] mramm: ah, home town [10:42] TheMue: home town now [10:43] rogpeppe: yea, still on copenhagen time [14:03] lunch === lynxman_ is now known as lynxman === TheMue_ is now known as TheMue === mcclurmc is now known as mcclurmc_away [16:00] * Aram curses in anger [16:02] * fwereade makes vague but soothing and conciliatory noises [16:04] is this known? [16:04] firewaller_test.go:712: [16:04] s.assertEnvironPorts(c, []state.Port{{"tcp", 80}}) [16:04] firewaller_test.go:71: [16:04] c.Fatalf("timed out: expected %q; got %q", expected, got) [16:04] ... Error: timed out: expected ["tcp:80"]; got ["tcp:80" "tcp:8080"] [16:07] meeh [16:07] of course the "simple" watcher niemeyer "fixed" because mine was too complex and submitted even when I said it wasn't good doesn't work correctly. [16:17] Aram, sorry, I haven't seen that one === mcclurmc_away is now known as mcclurmc [16:34] Aram: Interesting, I'm just looking. So far it worked. [16:37] Aram: Here it passes, just took a fresh branch of the current trunk. [16:37] yeah, it passes with trunk, it doesn [16:38] 't pass with the machine units watcher instead of the old machine principals units watcher [16:47] Aram: Strange ... [17:26] fwereade: around? [17:27] Aram, for a few moments, what can I do for you? [17:27] fwereade: the firewaller assumes the machine units watcher will report changes in unit port settings. [17:27] Aram, ha, interesting [17:28] it seems sensible to me to allow the machine units watcher to return any change in any attribute rather than only changes in lifecycle. [17:28] and the client can filter things out. [17:28] Aram, that seems like a sane alternative to one-watcher-per-unit [17:28] Aram, I think I'm +1 on that [17:28] cool. [17:28] Aram, I am well used to assuming that changes may not be relevant [17:28] Aram, the main thing is having everyone agree on what changes should be reported [17:28] yeah. [17:29] Aram, and honestly the "right" thing is irritatinglysituational [17:30] Aram, eg I *think* the uniter would work OK given extra service relations events but it was written against the current implementation [17:40] * rogpeppe has just emerged from an irritatingly long dive into the openssl source, just so i can work around the fact that mongod can't read an unencrypted PEM file. [18:36] * rogpeppe is off for the night. see y'all tomorrow === mcclurmc is now known as mcclurmc_away === otubo1 is now known as otubo === mcclurmc_away is now known as mcclurmc