[10:20] <xnox> I don't think i've setup ubuntu distribution correctly in my lxc launchpad
[10:21] <xnox> since it has no suites & thus no arches.
[10:23] <cjwatson> xnox: https://dev.launchpad.net/Soyuz/HowToUseSoyuzLocally
[10:23] <wgrant> xnox: The sampledata is... somewhat corrupt, ancient, incomplete, and otherwise undesirable.
[10:23] <cjwatson> Particularly soyuz-sampledata-setup
[10:23] <wgrant> xnox: There's utilities/soyuz-sampledata-setup.py to make it sane.
[10:23] <wgrant> Right, that.
[10:23] <cjwatson> That should be pretty much up to date - I followed that procedure when testing livefs
[10:23] <cjwatson> (and updated it slightly)
[10:26] <xnox> hm. so adding more debug it looks like it knows about releases/suites/pockets
[10:26] <xnox> it's just i don't have anything published =)
[10:26] <xnox> ./scripts/publish-distro.py -vv -C
[10:27] <cjwatson> FWIW, while there certainly are cases where I've wanted to actually run a local Launchpad instance and it's good not to have to do it in a rush, for the things I've been doing I've found it's better not to have it be my first resort for testing
[10:28] <cjwatson> (I couldn't have done all the livefs integration without it, but for the sorts of things I do that was the first time in about a year)
[10:29] <xnox> right. let me fake it. So i'm just copying a mock archive into /var/tmp/archive/ubuntu/ (e.g. an rsync of a smallish ppa)
[10:29] <wgrant> xnox: What're you trying to do?
[10:29] <xnox> which should be enough to run ./cronscripts/generate-contents-files.py
[10:29] <xnox> and it looks like it is =)
[10:30] <cjwatson> bin/test -vvct generate_contents_files
[10:31] <cjwatson> is at least slightly useful and you'll probably need to extend it anyway :)
[10:31] <xnox> yeah, i have run those already as well. But e.g. wanted to be able to run stuff by hand as well.
[10:31] <cjwatson> Fair enough
[10:32] <xnox> cause e.g. i want it to be a larges repository to benchmark improvements, if i can make any =)
[10:34] <cjwatson> Yeah, that might be worthwhile for the try-not-to-compute-everything-twice bug
[10:35] <cjwatson> BTW http://people.canonical.com/~ubuntu-archive/apt-mirror.cgi/1405620000/dists/utopic/main/binary-i386/Packages.bz2 (etc.) is now a link
[10:35] <cjwatson> *now a thing
[10:35] <xnox> =)))) nice
[10:35] <cjwatson> Hopefully it's not a complete security nightmare ... haven't hooked up any of the magic pool traversal stuff because don't have uwsgi etc. there and didn't want this to be an epic journey
[10:35] <cjwatson> So I cloned-and-hacked apt-mirror-web into a trivial CGI script instead
[10:36] <cjwatson> It'll 404 practically anything it doesn't like
[10:36] <xnox> cause go is for kids ;)
[10:37] <cjwatson> I only care about it for derive-distribution anyway :-)
[10:37] <xnox> right.
[10:38] <xnox> where is source for it?
[10:38] <cjwatson> hm, let me push it to github or something
[10:38] <cjwatson> *mutter*
[10:39] <xnox> i've checked people.canonical.com:~ubuntu-archive/public_html and it's obviously not there any more =)
[10:39] <cjwatson> xnox: https://github.com/cjwatson/apt-mirror/tree/ubuntu-archive
[10:40] <cjwatson> real web developers please try not to laugh too loudly
[10:40]  * wgrant plays pin the security hole on the CGI script.
[10:49] <cjwatson> Grr, bz2.BZ2File in 2.7 doesn't accept file objects, seriously?
[10:50] <cjwatson> hate Python 2
[10:57]  * xnox targeted python3 only....
[11:06] <cjwatson> this is for derive-distribution; don't want to rely on having your fancy new launchpadlib on my system by next week
[11:09] <xnox> =/
[11:09] <xnox> yeah, i think everything is in trusty to run apt-mirror, python3, python3-cherrypy, python3-uwsgi et.al.
[11:09] <xnox> doesn't use launchpadlib direct, just constructs restful url to download things.
[11:29]  * cjwatson considers derive-distribution's package selection
[11:30] <cjwatson> It occurs to me: I've got historical Sources/Packages, and I can get the seed output at a given time as well; I think the right answer is to do a time-travelled germinate run against a nominated list of seeds + optional extra packages
[11:30] <cjwatson> And then copy the closure of resulting sources
[11:30] <cjwatson> Possibly germinate on all relevant architectures, in fact, but that's fine, I have the technology
[11:32] <cjwatson> I can even make use of the never-used facility in germinate to plug in an alternative archive representation, which is basically perfect for this
[11:39] <cjwatson> Maybe double-check against actual image contents too
[15:51] <cjwatson> bzr's date revisionspec is irritating - no way to get the state of the branch *at* a particular date, you can only ask for the first revision after that date, and that breaks if the date in question is after the last commit on the branch
[15:52] <cjwatson> Will have to figure it out by hand I guess
[20:26] <james_w`> cjwatson: can you combine with before: to get the behaviour that you want?
[20:27] <james_w`> it might get you the state at that date, but probably changes the failure mode to when the first revision is after the date
[22:02] <cjwatson> james_w`: I've already done it differently, but before:date:... was where I started
[22:02] <cjwatson> It evidently tries to resolve date: first, fails to do so, and then gives up on the whole revisionspec; it makes sense in a piecewise way but isn't very convenient all round
[22:03] <cjwatson> I just did a crude "bzr log" parser instead and will no doubt pay the karma cost at some point.  This doesn't have to be fast or especially maintainable