[03:41] Bug #1812605 opened: snap find crached [06:07] morning [07:13] Hi [07:14] A bit in recovery mode [07:14] zyga: hey [07:14] Fell off the stairs last night [07:14] zyga: oi, are you ok? [07:14] I may join standup but don’t wait for me [07:15] When I can I will work on packaging [07:15] So so [07:15] zyga: was the trip back ok at least? :) [07:15] Some bruises and stuff like that. Some more back pain [07:16] Currently with wife for pregnancy checkup [07:16] Yeah, the trip was ok :-) [07:16] Just long and not much sleep on the way [08:09] PR snapd#6294 closed: packaging/ubuntu: build with golang 1.10 [08:12] yay [08:12] mvo: hey [08:15] PR snapd#6391 closed: tests: simplify interfaces-contacts-service test [08:16] hey mborzecki - good morning! [08:16] mborzecki: yeah, indeed yay :) [08:16] mvo: 1.9 is the new minimum now? [08:16] mborzecki: correct, we should probably update our readme too [08:17] mvo: hm there's no metion of any particular go version in the readme [08:18] mborzecki: yeah, I think we soon add a mention [08:18] mborzecki: its not OMGnow important though :) [08:18] mborzecki: also I just sent a mail out to foundations, security, sru etc to make sure they know we really did that, we did discuss it but I want to double check noone is unaware [08:19] geh, everyone is aware (double negatives are terrible!) [08:19] we can replace golang.org/x/net/context with context now [08:19] and update govendor too [08:21] mborzecki: yes, lets discuss in the standup. I would like to wait until we got feedback from foundations/security etc to make sure we don't need to revert anything. but having PRs will be good [08:22] i can do a quick cleanup and open a PR if we are greenlit to move forward with this [08:29] mborzecki: sounds great, please do [08:30] wow, we even used ctxhttp [08:37] 35 files changed, 36 insertions(+), 66 deletions(-) [08:37] not an awful lot [08:38] nice [08:40] golang.org/x/net/context/ctxhttp goes away, but x/net/context is imported by tomb, so it has to stay around [08:41] mborzecki: ok [08:58] we could also drop the check for numerical http return codes [08:58] and use consts from net/http [09:02] PR snapd#6402 opened: spread: increase default kill-timeout to 30min [09:09] morning peeps [09:12] Chipaca: hey, good morning! [09:12] Chipaca: morning [09:17] PR snapd#6403 opened: many: cleanup golang.org/x/net/context [09:34] mborzecki: nice, thank you! [09:34] mborzecki: looks like unit tests are unhappy in this PR, seems like new stuff that the new go checks [09:34] mvo: mhm, pushing in a minute [09:34] mborzecki: also only in tests so shouldn't be bad :) [09:34] mborzecki: no worries, just wanted to let you know [09:34] * mvo is quite happy about this PR [09:45] mvo: and pushed [09:49] ta [10:11] * Chipaca removes 1.6, installs 1.10 [10:11] * Chipaca feels like it's a party [10:12] * mborzecki can drop 1.6 gofmt wrapper now [10:15] oooh, there's a bunch of things we can drop [10:15] Chipaca: can you take a look at #6403 ? [10:15] PR #6403: many: cleanup golang.org/x/net/context [10:16] mborzecki: egrep -r --exclude-dir vendor '// \+build \!?go' [10:17] mhm [10:17] I need to double-check but IIRC ctrl17 was good up to 1.10 at least [10:17] Chipaca: was there some fancy json decoding too? [10:17] 1.11 might be using an even newer unicode but ctrl didn't see much change [10:18] mborzecki: ctrl is part of that, if you mean the 'clean' json [10:18] but [10:18] mborzecki: there are places where we use *json.RawMessage instead of json.RawMessage, that we could drop [10:18] and see some perf improvement [10:18] perhaps [10:18] pr by pr :P [10:20] ep [10:30] had to get https://packages.ubuntu.com/bionic/amd64/golang-golang-x-tools by hand to get godoc to play nice again [10:30] (i'm on xenial here) [10:31] * sparkiegeek wonders if Chipaca could have used https://launchpad.net/adapt [10:33] sparkiegeek: no idea what that is :-) [10:34] ah [10:34] sparkiegeek: no need to rebuild the package, it just works as is [10:35] Chipaca: it installs Ubuntu packages from 'other' series in a LXD and wraps them to make a cheap way of getting package X available from a given series [10:35] * Chipaca imagines using X from bionic o n xenial, and shudders [10:36] hah :) [11:19] o/ [11:19] hey zyga [11:19] hey mvo :) [11:19] zyga: welcome back! [11:20] zyga: but please take it easy today [11:20] mvo: I'm sorry about expensify, it's super easy to click the wrong button while on the go [11:20] (about that "let me expense one coffee thing") [11:20] I'm sorting out my travel topics now [11:20] and recovering after the small accident yesterday :) [11:20] but all will be fine [11:21] anything urgent, any news? [11:22] zyga: no fires [11:22] zyga: your go-1.10 pr is in \o/ [11:23] I noticed, thank you, good idea to notify other teams about that [11:49] zyga: looking into bcond_with{,out}, do we have 15.0 image for spread? [11:51] mborzecki: mmm, not sure [11:51] if you have a branch I have all the VMs ready [11:51] oh [11:51] one thing [11:51] so [11:51] tl;dr; I spent last week working on top of ubuntu in hyper-v and multipass also on top of hyperv [11:52] mborzecki: I think we could start standardizing on using multipass for test-builds on native kernel [11:52] it's very flexible and qucick [11:52] quick* [11:52] does it work well with non ubuntu vms? [11:52] yes [11:52] only requirement is cloud-init in the image [11:52] I've got a pipeline for debian builds now [11:53] http://download.opensuse.org/repositories/Cloud:/Images:/Leap_42.3/images/ [11:53] I will give this a try (though not with any priority today) [11:54] mborzecki: it's as simple as "multipass launch $URL-to-qcow2 -c 10 -m 4G -n vm-name [11:54] then you have multipass commands to exec stuff inside, copy stuff around, shell interactively [11:54] multipass handles networking and caching [11:54] ah, haven't tried anything else than ubuntu with multipass yet [11:55] what if the image has no cloud-init? [11:55] nogo [11:55] I think it's a requirement [11:55] ask Saviq once he's back [11:55] I'm here [11:55] Saviq: is cloud-init a hard requirement? [11:56] yes, otherwise we can't access the launched instance [11:56] Saviq: did I say thank you for multipass, it's awesome [11:56] mborzecki: otherwise you'd need to know (and pass it to us) a default username/password [11:57] and we don't support that right now, nor look forward to it really [11:57] mborzecki: does arch have any cloud images? [11:57] everything else we support has them [11:58] anything else, we'd need to expose a graphical and/or serial console [11:58] PR snapd#6404 opened: snapstate: auto transition on experimental.snapd-snap=true [11:58] and it'd be a hands-free experience from then as far as multipass is concerned [11:58] zyga: nah :P no cloud images afaik [11:58] you can probably build one [11:58] well, we built one for spread [11:58] but it's not using cloud-init afaik [11:59] probably has a well-know root password :) [11:59] +n [11:59] yup, something no trivial, like 123 :P [11:59] https://wiki.archlinux.org/index.php/Cloud-init [12:00] https://wiki.archlinux.org/index.php/Arch_Linux_AMIs_for_Amazon_Web_Services [12:00] may very well be possible to use those [12:01] mh [12:02] Saviq: while at it, is it possible to launch throw-away instances with multipass? i often work directly with cloud images and use qemu .. -snapshot [12:04] mborzecki: we have a handful of approaches planned for speeding up boot, but we're unlikely to add full snapshot support any time soon, don't want to become just another CLI for hypervisors [12:13] PR snapd#6405 opened: run-checks: ensure we use go-1.10 if available [13:35] mvo: mount error still happens [13:35] error: cannot perform the following tasks: [13:35] - Mount snap "network-consumer" (unset) ([start snap-network\x2dconsumer-x1.mount] failed with exit status 1: Job for snap-network\x2dconsumer-x1.mount failed. [13:35] See "systemctl status "snap-network\\x2dconsumer-x1.mount"" and "journalctl -xe" for details. [13:35] ) [13:35] this is from the PR above [13:46] zyga: is the go version relevant enough to be in 'snap version'? [13:46] Chipaca: dunno [13:46] perhaps? [13:46] zyga: or would this be a case for a --verbose flag or sth? [13:46] yeah [13:46] or even a debug [13:46] snap debug build-info [13:46] i mean, if it's for the tests, just add a 'go version' :-) [13:46] could show other helpful stuff [13:46] +1 [13:46] go version can be elusive :) [13:46] if we have many [13:51] ooh, ooh! I remembered what I wanted to do when we moved past 1.7! [13:52] https://gist.github.com/chipaca/10ecdf364db962c9b5ee590f0f610b2a [14:08] sorry for cutting me video feed but I'm out of power on the phone [14:13] zyga: we lost you [14:22] zyga: yeah, cachio mentioned this as well, on what system did that happen for you? [14:32] mvo: that was in the PR you sent, I don't recall [14:32] mvo: I only saw it there [14:38] small helper I use for packaging work [14:38] packaigng helper https://www.irccloud.com/pastebin/oPwDE5pt/ [14:38] mvo: ^ [14:39] mborzecki: ^ [14:41] hm we should add a multipass backend to spread :) [14:43] PR # closed: snapd#5644, snapd#5822, snapd#5962, snapd#6016, snapd#6034, snapd#6079, snapd#6098, snapd#6106, snapd#6108, snapd#6111, snapd#6162, snapd#6177, snapd#6238, snapd#6252, snapd#6258, snapd#6270, snapd#6280, snapd#6281, snapd#6313, snapd#6320, snapd#6322, snapd#6324, snapd#6325, [14:43] snapd#6327, snapd#6329, snapd#6333, snapd#6341, snapd#6347, snapd#6348, snapd#6356, snapd#6360, snapd#6363, snapd#6367, snapd#6376, snapd#6380, snapd#6381, snapd#6383, snapd#6387, snapd#6389, snapd#6394, snapd#6396, snapd#6400, snapd#6401, snapd#6402, snapd#6403, snapd#6404, snapd#6405 [14:43] PR # closed: core-build#11, core-build#22, core-build#26, core-build#37 [15:07] mvo, should I remove the manual from debian sid after the imagge is updated? [15:07] cachio: yeah, lets try that [15:07] cachio: once its updated, lets push a PR that enables sid again [15:07] mvo, sure, thanks [15:07] cachio: and then we can check how well that goes - it will be a good prep for the packaging merge PR. thank you! === sgclark is now known as sgmoore [15:54] * cachio lunch [16:28] popey, ping [16:28] om26er: hi [16:28] popey, could you take a look at this one https://github.com/asciinema/asciinema-snap/issues/1#issuecomment-454529232 [16:28] will do [16:42] PR snapd#6406 opened: tests: enable debian sid as part of the main suite on travis [16:43] my debian foo has grown the SBUILD_SHELL trick [16:53] I uploaded new test-snapd-dbus-{consumer,producer} snaps to edge, if tests suddenly explode, this is why [16:56] (the PR below addresses this) [16:56] PR snapd#6407 opened: tests: get test-snapd-dbus-{provider,consumer} from the beta channel [17:01] PR snapd#6408 opened: tests: add spread test for system dbus interface [17:41] wooot [17:41] I solved the debian unit test failures :) [18:56] zyga, this is the error you have fixed? https://paste.ubuntu.com/p/cmQTKpDhh6/ [18:56] no [18:57] this is a known issue about go fmt changing behavior across go versions [19:00] zyga, ahh, ok [21:01] Bug #1812751 opened: Several remove event for unknown device