[01:09] <wgrant> mwhudson: Which variant, and with which arguments?
[01:11] <mwhudson> wgrant: just getBuildRecords(pocket='Release') on a ppa
[01:11] <mwhudson> the ppa has 6000 builds which is quite a lot i guess, but still
[01:12] <mwhudson> it takes like 30 minutes
[01:13] <wgrant> mwhudson: Which PPA, and how are you using the call?
[01:13] <wgrant> 100 batches in 30 minutes is impossible; timeouts prevent it.
[01:14] <mwhudson> wgrant: https://launchpad.net/~mwhudson/+archive/ubuntu/devirt/+packages
[01:14] <wgrant> It's possible that the requests are timing out and launchpadlib is transparently retrying, I suppose.
[01:14] <mwhudson> ah that might explain it
[01:14] <mwhudson> you can make launchpadlib log requests can't you?
[01:15] <mwhudson> wgrant: it's this script https://git.launchpad.net/~barry/+git/pydeps/tree/upload.py
[01:20] <wgrant> mwhudson: In [4]: t = time.time(); bs = list(a.getBuildRecords(pocket='Release')); print time.time() - t
[01:20] <wgrant> 259.202159882
[01:20] <mwhudson> i just killed the script after 32 minutes :(
[01:21] <wgrant> mwhudson: Oh, the script as a whole takes more than 30 minutes? That's more plausible.
[01:21] <wgrant> mwhudson: Also that doesn't do getBuildRecords(pocket='Release'). It does getBuildRecords(source_name=name) for many names.
[01:21] <wgrant> And it's also presumably designed to run from the US, not APAC.
[01:22] <wgrant> You'll probably find that the script is doing several requests per package.
[01:23] <mwhudson> that's not the bit that's the problem
[01:23] <mwhudson> i have now yes :(
[01:23] <mwhudson> is there any better way of doing this?
[01:23] <mwhudson> screen scraping the ppa listing would be faster but ugh
[01:25] <wgrant> Which bit is the problem?
[01:29] <mwhudson> well the goal of the script is just to a list of source package names and build status
[01:33] <wgrant> mwhudson: getPublishedSources and getBuildStatusSummariesForSourceIds (or just a global getBuildRecords call, since you presumably want everything) should be able to do it quickly enough.
[01:34] <mwhudson> wgrant: does launchpadlib cache intelligently enough that if i call list(getPublishedSources()), build.current_source_publication won't hit the network each time?
[01:36] <wgrant> mwhudson: No.
[01:36] <mwhudson> sigh
[01:36] <wgrant> Correct.
[01:37] <wgrant> This is why pure hardcore REST APIs are very silly.
[01:39] <mwhudson> does launchpad have any sql injection vulnerabilities i can use for this? :)
[02:03] <mwhudson> it's much faster run from canonistack :)
[02:10] <mwhudson> wgrant: is it known that the headless login process with launchpadlib is kinda busted
[02:10] <mwhudson> ?
[02:14] <wgrant> mwhudson: Yes, but it works OK as long as you don't try to use the autolaunched web browser.
[02:15] <mwhudson> wgrant: i meant this http://paste.ubuntu.com/24534363/
[02:15] <wgrant> mwhudson: Huh, never seen that before.
[02:19] <mwhudson> wgrant: happens on artful, haven't checked anything older
[02:20] <mwhudson> oh yes and if i fix that this happens: http://paste.ubuntu.com/24534385/
[02:20] <mwhudson> (the fix for that is to log in with python 2 first ...)
[02:23] <wgrant> mwhudson: Oh, Python 3 launchpadlib isn't really a thing
[02:23] <wgrant> It's meant to mostly work but I've never used it.
[02:23] <mwhudson> wgrant: i see
[02:27] <mwhudson> ah forcing unecrypted creds works ok
[11:46] <redwolf> o/
[11:47] <redwolf> I need a bit of help with the @ubuntu alias
[11:48] <cjwatson> redwolf: What's up?
[11:48] <redwolf> I had to change my username, everywhere (launchpad, SSO, wiki, etc), added a gmail account matching username
[11:49] <redwolf> I get this message: http://pastebin.ubuntu.com/24536346/
[11:49] <redwolf> is that because it needs time?
[11:50] <cjwatson> redwolf: OK, you'll need to ask #canonical-sysadmin about that (or rt@ubuntu.com if you don't get a response there).  Launchpad holds the data, but it's our sysadmins who actually do the processing into email aliases.
[11:50] <redwolf> ok. I will. thanks anyway :)
[11:51] <cjwatson> I believe there's some manual batch processing involved.
[11:51] <redwolf> I think so too. flushing database or something
[11:52] <redwolf> ta! o/