[07:48] <RAOF> alan_g: Mir KMS tests fail on 18.04 because drmModeAddFB2 had changed its signature.
[07:48] <RAOF> (and others, but that's what's causing the initial crashes)
[07:49] <RAOF> Added const to the array parameters.
[07:49] <RAOF> In case Saviq asks 😀
[08:52] <Saviq> ohkay and why did bors merge it when I did bors r- ¿?
[09:05] <alan_g> I don't know. But then I'm new to bors.
[23:15] <Saviq> RAOF: if you're not thanksgiving, I'd give thanks for https://github.com/MirServer/mir/pull/48 ;)
[23:15] <RAOF> Just got the email notification :)
[23:15] <Saviq> RAOF: also, any idea why this got merged even after I "bors r-"'d it?
[23:16] <RAOF> No, I don't.
[23:16] <RAOF> I'll have a look.
[23:18] <Saviq> RAOF: UGH, seems I got tricked by git again, the rev-parse thing has the same problem of not looking at the remote, only at what's fetched
[23:18] <RAOF> Hah.
[23:19]  * Saviq feels we should use my dumbcounter until we can use `git describe` :P
[23:20] <RAOF> Will `git describe` have the same problem?
[23:21] <Saviq> I think it might if the closest tag is far enough in the history
[23:22]  * RAOF again curses git for not having a revno.
[23:22] <Saviq>  git describe --tags
[23:22] <Saviq> fatal: No names found, cannot describe anything.
[23:23] <Saviq>  git fetch --deepen 100
[23:23] <Saviq>  git describe --tags
[23:23] <Saviq> v0.27.0-641-g766a4f39d
[23:23] <Saviq> so one thing we could do here is `git fetch deepen 100` until we get a description out of `git describe`
[23:24] <RAOF> Or we could just fetch the whole history.
[23:25] <RAOF> It's not that big.
[23:27] <Saviq> 160MB
[23:28] <Saviq> .5s to fetch --depth=50 (what travis does) vs. 16s for a full clone
[23:28] <Saviq> I'd be happier to fetch back 100 at a time until we get `git describe` to work
[23:28] <Saviq> again, if git tags are what we want to use for this...
[23:32] <RAOF> Our test suite takes at least 600s. Is saving 16s for not doing a full clone worth it? :)
[23:34] <Saviq> RAOF: maybe, maybe not ;)
[23:34] <Saviq> ok for now I'll just go --unshallow
[23:35] <Saviq> so we can fix this
[23:35] <RAOF> Yeah.
[23:35] <RAOF> Launchpad doesn't (yet) allow us to webook successful PPA builds into a GitHub status, right?
[23:38] <Saviq> RAOF: it can do autopkgtests
[23:38] <Saviq> where from/how, I'm not sure
[23:39] <RAOF> Because that would be a good thing to gate bors on…
[23:39] <RAOF> It would avoid the Travis timeout.
[23:39] <Saviq> agreed
[23:40] <Saviq> depending on what the autopkgtests hook does (I can't see results on snapd's PRs any more)
[23:41] <Saviq> any case, we'd need to create ephemeral PPAs per PR for this to make sense (but yeah, we should!)
[23:41] <Saviq> ooh my first signed commit, /me likes the green :P
[23:42] <RAOF> Why would we need ephemeral PPAs?
[23:42] <RAOF> Oh, yeah.
[23:42] <RAOF> Never mind :)
[23:42] <Saviq> RAOF: yeah, that :)
[23:42] <Saviq> RAOF: ok, https://github.com/MirServer/mir/pull/48/ should be good for real this time
[23:44] <Saviq> oh and btw, we got the bot key in https://launchpad.net/~mir-team/+archive/ubuntu/dev/+packages
[23:45] <Saviq> so after that lands we'll have staging populated