[02:17] * thumper takes kid to sports === thumper is now known as thumper-afk [02:29] wallyworld: I'm probably not going to make our 1:1 today, my son has his "Winter Musical" this morning [02:29] ok [02:29] enjoy :-) === thumper-afk is now known as thumper [04:05] thumper: hey. i added fingerprint support to the ssh utils [04:05] wallyworld: cool [04:05] had to call out to ssh-keygen [04:06] i couldn't find a Go implementation === thumper is now known as thumper-afk === thumper-afk is now known as thumper [07:44] jam: I'm actually around now [07:45] if fwereade_ is too [07:45] thumper: I'm around as well, not sure about fwereade_ [07:45] I had another call at 0800UTC, but it has been moved [07:47] thumper: well, william responded to one of my emails at 11pm last night, so it may be that he's going to be a bit later today [07:54] thumper: don't forget to look at my ssh branch again :-) tomorrow first thing your time will be fine :-) [07:57] wallyworld: ack [07:58] thanks [08:00] jam, thumper: if you're both here let's chat now [08:07] fwereade_: I'm just chatting with dimiter for a bit, and then I'll be right with you guys [08:12] sorry, was making a coffee [08:12] fwereade_: , jam (when you're ready): https://plus.google.com/hangouts/_/7ecpitcov2peobbk7u31s52nn0?hl=en === rogpeppe1 is now known as rogpeppe [09:03] mornin' all [09:06] hey rogpeppe, just in another call, will be there in a sec [09:16] rogpeppe: I'm in now, whenever you're available [09:58] mgz: good morning! [09:59] I'm just about to go make a coffee, and then I'll be around, mumble or hangout is fine for me. [09:59] jam: sure [10:10] mgz: I'm back, mumble or hangout? [10:10] jam: mumble! [10:11] mgz: I'm already there [10:11] mgz: you're completely unintelligible [10:11] it's unhappy... [10:11] he-e-e-l-l-l-o-o-o-o-o [10:12] that's how you're coming through for me [10:45] time to stand up, raisin' the roof [10:45] https://plus.google.com/hangouts/_/calendar/am9obi5tZWluZWxAY2Fub25pY2FsLmNvbQ.mf0d8r5pfb44m16v9b2n5i29ig [10:45] rogpeppe: fwereade_, TheMue ^^ [10:45] dimitern: ^^ [10:48] can someone give me the meeting link? my chrome is acting up [10:48] natefinch: https://plus.google.com/hangouts/_/calendar/am9obi5tZWluZWxAY2Fub25pY2FsLmNvbQ.mf0d8r5pfb44m16v9b2n5i29ig [11:42] fwereade_: for later when you are free, i renamed "credentials" package to "keyupdater" https://codereview.appspot.com/37570043 [12:29] dstroppa: dimitern mentioned you want to run `go fmt ./...` on lp:gojoyent [12:29] and the others are also taking a look at the cl [12:30] thanks, will do that [12:30] if we did it all correctly, you should be able to make changes then run the same lbox command to update the review [12:31] dstroppa, mgz, thanks! [12:58] jam: i updated the firmware on my router and it seems to have fixed the issue [12:59] jam: (even though the firmware update notes contained no mention of the bug) === gary_pos` is now known as gary_poster [13:31] rogpeppe, jam, fwereade_, mgz, et al - just letting you know ^^ [13:31] dimitern_: just letting us know what? [13:31] ah, sorry [13:31] * dimitern_ needs to be off for a while, the electrician came to change the wires and i won't have power for at least 1 hour [13:32] ( though I managed to send it, but apparently not) [13:32] dimitern: good luck :) [13:32] mgz, cheers :) === wedgwood_ is now known as wedgwood === smoser` is now known as smoser === sparkieg` is now known as sparkiegeek [17:10] rogpeppe: query about doing providery things from inside the apiserver [17:10] mgz: go on [17:11] to actually recreate status as it exists currently, I'd need to do provider Instances() to get instance statuses [17:11] this is something the apiserver doesn't actually want to do... how can I get around that for now without borking things? [17:11] mgz: if we can avoid that, we should [17:12] currently the output is just different... [17:12] mgz: eventually, we probably want a worker that keeps the instance status in the state up to date [17:12] mgz: for the time being, perhaps you can't avoid it [17:12] one option is to leave some of that client-side for now [17:12] mgz: let's not do that [17:13] but that makes api-status less fun [17:13] mgz: that loses a lot of the bonus of client-side status [17:13] s/client-side/API-based/ [17:13] right [17:14] so, it's not completely unsafe to construct and use a provider inside an api call? I note some make one to look at config stuff [17:14] mgz: no, it's notr [17:14] mgz: and some calls currently do that, i think [17:14] but nothing seems to actually use the provider to do and cloud-operations, so I'm a little scared [17:15] mgz: well, currently the deploy API call (whatever it's called) does [17:15] mgz: it puts stuff into provider storage [17:15] ah, I'll have a look at that, a grep didn't turn it up [17:15] mgz: look for ConnFromState [17:16] mgz: yeah, looks like it's just charm putting [17:17] mgz: i think that for the time being, we should probably just move the provider Instance calls into the API. [17:17] rogpeppe: okay, thanks, I'll have a play around [17:18] mgz: but... [17:18] mgz: if we can, we should avoid iterating when the instance doesn't exist [17:18] mgz: and we should make only one provider call if possible to get all the instances [17:19] mgz: that way the overhead's not too bad [17:19] well, my current scheme is just to bung everything behind one single call [17:19] not split the instance fetching and machine fetching at all [17:20] mgz: i'm talking about the implementation of the Client.Status call [17:21] mgz: ah, looking at the current implementation, that's what it does anyway, [17:21] mgz: so that's fine [17:22] right, in the cmd status it just lists everything once then sticks in in a struct to do further work on [17:24] mgz: yeah - that's just fine [17:25] mgz: the place we want to be a bit more intelligent is in the address updater, but that's a different discussion [18:07] jam, fwereade_ re the manual provider support in 1.16.. does that effect the api ongoing for 1.18/trunk? [18:41] * rogpeppe is done for the day [18:45] hazmat: inasmuch as we want to be backwards compatible to what we had in 1.16 (fallback to direct DB access), it does, but I don't think it is massively deep. A couple hours for me to finish making 1.17 compatible with the 1.16 manual provisioning steps. [18:45] * jam is off to bed [18:46] night jam! [18:48] jam.. ic. so no api changes per se to 1.18/trunk? [18:48] jam, g'night.. i just want to avoid non go clients from contemplating tool lookup, the current machine config api does that nicely. [22:20] any one know if the result from os.Pipe() is serializable over net/rpc ? [22:21] just considering options in testing... [22:21] but I suppose I can test it now... [22:21] * thumper pokes