=== ogasawara_ is now known as ogasawara [08:48] Laney: flushed another pile of Haskell stuff through NEW === yofel_ is now known as yofel [11:13] cjwatson: cheers, http://orangesquash.org.uk/~laney/transitions/ghc.html doesn't look so bad [11:13] although... [11:13] just noticed that https://launchpad.net/ubuntu/+source/haskell-hlist didn't get process-removals removed [11:13] did the buildX mess it up? [11:25] not sure, slangasek was handling process-removals [11:26] I think only an 'ubuntu' substring is meant to inhibit p-r [11:35] * Laney goes blind at the p-r code :-) [11:39] http://ftp-master.debian.org/removals.822 does exist now [11:51] Laney: removed now [11:54] and haskell-hgl [11:55] how come stuff isn't going green on the ghc tracker, but only to unknown? [11:56] oh, wait, it is, I'm stupid [12:18] lamont: so, I'm looking at converting livecd-rootfs to live-build, and trying to figure out what would be involved [12:19] * lamont has no knowledge on live-build [12:19] lamont: from your POV, would it be an acceptable interface for buildlive to call ssh $buildd BuildLiveCD , and then BuildLiveCD could call lb config then lb build? [12:19] (which is AIUI basically the normal sequence) [12:19] lb config has vast swathes of possible options [12:20] where does the chroot command live in that pile? [12:20] so we should be able to add options until it's equivalent to livecd-rootfs [12:21] I assume we'd have BuildLiveCD call lb * within a chroot, much as it currently calls livecd.sh [12:21] lb can optionally do its own chrooting, but I don't think that's useful for us [12:21] since we'll need to pick the right version of live-build anyway [14:02] Laney: I guess the ghc tracker is failing to update - do you know why? [14:02] cjwatson: I haven't cronned it yet [14:02] aha :-) [14:02] shamefully it was just a big && line [14:02] I'll write a proper script now that there's more than one thing to track [14:03] is any assistance needed with uploads, or is it all just in a giant dep-wait stack? [14:04] (NEW activity suggests the latter ...) [14:04] hopefully it should just be dep-waits [14:05] i'll need to do some merges/sync requests [14:05] and then if my account gets created soon I'll upload the remaining packages to debian [14:06] so from your side just keeping an eye on NEW should be enough [14:06] ok [14:08] removing haskell-network-bytestring [14:09] hmm, actually maybe I should leave that until haskell-network has built everywhere [14:13] which is ultimately down to haskell-transformers failing, which looks transient - I see you just retried that? [14:13] yep [14:14] any idea why haskell-transformers didn't show up on the tracker, though? it looks as though it should have done [14:15] perhaps it looked for binaries which matched the source version [14:15] it showed up as 'good' — could be a bit of a hole [14:16] 43 sync requests outstanding, sigh [14:17] uds hangover? [14:18] yeah [14:21] cjwatson: I suspect that BuildLiveCD wants to take a distroname, which it converts to a chroot name, and then chroots into that to do everything thereafter [14:21] but yeah, sounds about as sane as anything else [14:28] lamont: I guess I didn't want to have significant amounts of configuration encoded in BuildLiveCD, because then I either have to create a package for it (or keep livecd-rootfs around indefinitely), or we end up with the configuration hard to access by platform [14:29] which is why I'd rather have the lb config args passed in from somewhere else [14:29] if that makes sense [14:29] yeah [14:30] I concur with having a trivially simple BuildLiveCD script - possibly even thrown into live-build, 'tever [14:30] we'll probably want our own for a while because it'll need to handle both livecd-rootfs and live-build [14:30] yeah [14:31] I could make such a change in livecd-rootfs and then keep it around until we stop needing livefs builds for I guess [14:31] BuildLiveCD is actually delivered outside of the package these days [14:31] yeah, though copied from the package, right? [14:31] yes [14:31] copied from the package into $magicplace, from whence it is blatted out to all the builders [14:32] so as it sits, you get exactly one version of BuildLiveCD, consistently, within about 10 minutes from when I push the change live [14:38] zomg arm might catch up this week [15:40] lamont: yay [16:17] * ScottK wonders if he should upload KDE then. [16:20] cjwatson, Laney: no clue why haskell-hlist didn't get caught in my last p-r run [23:05] Laney: in ghc.ben, shouldn't several of the * characters in your regexes be .* ? [23:36] are there any pre-alpha builds available? [23:41] no CD image-type builds, no [23:41] that's probably one of the tasks for this week [23:41] k [23:46] cjwatson: yes :-)