[00:46] wgrant: Do I have to drop indicies that reference a column before dropping it or will Postgres just deal? [00:46] StevenK: postgres will drop any indices that involve the column [00:46] StevenK: This can sometimes be undesirable, in the case of compound indices [00:46] So be careful [00:47] product is fine, just a partial index, no compounds [00:48] distribution is also fine, one index, no compounds [01:10] StevenK: Right [01:20] wgrant: make newsampledata seems to ignore the column drop [01:21] StevenK: You've applied the patch to launchpad_dev and launchpad_ftest_playground? [01:23] Hmm, my apply ... didn't. [01:23] * StevenK stabs it harder [01:30] StevenK: Fail [01:30] StevenK: Instead of subscribing the security contact or the maintainer, you now unconditionally subscribe the maintainer [01:30] That block should go [01:30] Both of them [01:31] 1121+ getattr(info, 'bug_supervisor_tasks').append(value) [01:31] That can be simplified [01:35] wgrant: http://pastebin.ubuntu.com/1160033/ [01:36] what the driver [01:36] Oh [01:36] That's during the transition [01:36] Hm [01:38] I think that diff is correct [01:38] But I was never a fan of the unsubscription rules [01:43] wgrant: Thanks, I've commited and pushed that bit [02:12] * StevenK stabs the branch scanner for being terrible [02:32] Dear branch scanner, DIAF. [02:47] wgrant: Is https://code.launchpad.net/~stevenk/launchpad/db-destroy-security-contact cursed? [02:48] StevenK: Let me consult with my colleagues [02:49] StevenK: The spirits have indeed placed a curse upon it [02:49] Rename and repush [02:56] * StevenK sacrifies wallyworld to the branch scanner. [02:56] no not me, i've also had the same issues [02:56] so i vote to sacrifice wgrant [02:56] He's useful to have around. [02:57] but he's a young virgin and the gods like those [02:57] ROFL [02:57] quotes page :-D [02:58] wallyworld: It looks like threatning to sacrifice you was enough [02:58] yay [02:59] Heh [02:59] StevenK: Did you create an early MP or subscription or anything on that branch? [03:00] Before it had first scanned? [03:00] wgrant: Nope [03:00] :( [03:00] I didn't touch it for fear of a curse. [03:00] And it got cursed anyway [04:29] wgrant: Hahahaha [04:30] wgrant: You added POLICY_ALLOWED_TYPES to code/model/branchnamespace.py for branches and bugs/interfaces/bugtarget.py for bugs. [04:31] StevenK: Is that a problem? [04:31] Their meanings are different [04:31] And they were private until you came along [04:33] wgrant: I was amused at the model vs interface split [04:33] Ah, right [04:33] I wasn't quite sure where to put the latter [04:33] Since it's not used by model.bugtarget [04:33] But by the things that inherit it [04:34] Whereas for BranchNamespace it's only used by that class and tests. [04:34] And now a garbo job === almaisan-away is now known as al-maisan [05:24] wallyworld_, wgrant: https://code.launchpad.net/~stevenk/launchpad/remove-unused-sharing-garbo/+merge/120698 [05:26] StevenK: The RAM, oh god, the RAM [05:26] StevenK: just running out for school pickup, can look when i get back [05:44] wgrant: What do you think we should do to get the used or even better, the unused APs? [05:45] StevenK: You could search for artifacts for each and use is_empty [05:45] Means a couple of extra queries [05:45] But meh [05:55] Isn't apas = getUtility(IAccessPolicyArtifactSource).findByPolicy( [05:55] Already doing so? [05:56] And looping over it is bad? [05:58] StevenK: That's finding all the APAs, not just the APAs' distinct policies [05:58] launchpad_dogfood=# SELECT policy, COUNT(*) FROM accesspolicyartifact GROUP BY policy ORDER BY COUNT(*) DESC LIMIT 20; [05:58] policy | count [05:58] --------+------- [05:58] 137 | 45598 [05:58] 68775 | 10233 [05:59] Yeah, I see what you mean. [06:01] wallyworld_, wgrant: https://code.launchpad.net/~stevenk/launchpad/moar-auditor/+merge/120706 [06:02] StevenK: Why's lp.services.auditor.server not in testing/fixtures/something? [06:02] I was following txlongpoll's pattern [06:10] StevenK: Antipattern [06:11] Still [06:21] wgrant: Back to the garbo job: http://pastebin.ubuntu.com/1160294/ [06:26] unused_aps = [ap for ap in candidate_aps if apa_source.findByPolicy([ap]).is_empty()]? [06:38] wgrant: http://pastebin.ubuntu.com/1160307/ [06:39] StevenK: I'd tend to wrap just before the 'if' for cleanliness [06:41] wgrant: http://pastebin.ubuntu.com/1160309/ [06:42] StevenK: Also, unless the two args to both get() calls don't fit on one line, please wrap before the first arg. But apart from that it looks excellent. [06:42] And I'm glad access is no longer policing. [06:43] wgrant: http://pastebin.ubuntu.com/1160312/ [06:43] Yes, access made constable. [06:57] wgrant: No opinion? [07:00] StevenK: Looks good [07:01] wgrant: Thanks, pushing it up. [07:21] wgrant: Diff *finally* updated [07:26] good morning [07:33] wgrant: No approval for me? :-( [07:36] StevenK: Could you a docstring to the TunableLoop and perhaps some comments describing what it's doing and why? [07:36] Apart from that it's fine [07:37] I accidentally the whole thing, but anyway === mthaddon` is now known as mthaddon [08:27] Spads: Hi. Since the DC move, we've been short all the dogfood.lp builders, and rather a lot of Launchpad revisions are now blocked on a single revision that requires dogfood build QA. There's been some insinuation that we might be able to get king for a while, but I'm assuming that's non-trivial since it hasn't happened yet; is there any chance we might be able to temporarily steal a production PPA builder for the ... [08:27] ... task? It should only take maybe half an hour. [08:28] Err, why did that end up on #lp-dev [08:28] Given that Spads isn't here [08:28] * cjwatson moves to a cleverer channel [12:35] is something wrong with qastaging? i get timeouts (OOPS-64204c86af12c8a1d8c6fa597795a531) and oopses (OOPS-307f7eabfe6fed73cfa46b6dbb3b21d5) all the time [12:36] BjornT: the db on there isn't primed up with requests and it times out a lot more often [12:36] BjornT: are you trying to qa something or just tinker with disposable data? There are some smaller projects that tend to work ok for testing purposes [12:37] BjornT: Project group bug listings on qastaging... good luck with that [12:37] They're slow on production [12:37] And not indexed well [12:37] I'd try something else, or a very very small project group :) [12:38] rick_h_: i'm trying out an api script of mine. i guess i could try on a smaller project. although, i get oopses just looking at bugs as well (OOPS-6bdbc2384d11ecd88e64f635d5162828) [12:39] Hm [12:39] BjornT: Oh, that one's an OOPS, not a timeout [12:39] qastaging lacks a DB patch [12:40] wgrant: any update on the DB patch on qastaging ? === abentley changed the topic of #launchpad-dev to: http://dev.launchpad.net/ | On call reviewer: abentley | Firefighting: - | Critical bugs: 4.0*10^2 [13:28] abentley: ping [13:28] rick_h_: HI. [13:44] adeuring, abentley, rick_h_ -- hey there. Everyone good to join now? [13:45] deryck: rgr [13:45] deryck: sure [13:45] deryck: yup === al-maisan is now known as almaisan-away === cody-somerville_ is now known as cody-somerville === cody-somerville_ is now known as cody-somerville [15:31] jcsackett: I am having terrible connectivity issues [15:31] jcsackett: can you rollback cjwatson's rev that blocks deployment: http://lpqateam.canonical.com/qa-reports/deployment-stable.html [15:34] BjornT: qastaging has had the missing column added, so it should have stopped OOPSing. But I'm afraid there'll still be lots of timeouts [15:36] Hey, does the Launchpad code include a PPA picker in the same way that it has a person picker and a project picker? [15:37] no [15:37] mpt: no ppa-picker [15:37] There is actually a PPA picker [15:37] It's very basic [15:38] roe recipes? [15:38] for [15:38] Precisely. [15:38] sorry, network and stress make my only tpy in Curtisese [15:38] The "Daily build archive" field on SourcePackageRecipe:+index uses it [15:40] wgrant, basic in what way? That it lets you choose from only your own PPAs? [15:41] mpt: And your teams' [15:41] mpt: Without searching [15:41] Is that the same as the terrible list drop-down on +copy-packages? [15:41] cjwatson: It's arguably worse [15:41] although it at least shows the actual name [15:41] Not just the displayname [15:42] But it's paginated [15:42] (Which is ambiguous if displaynames ever collide across teams) [15:42] And unsearchable [15:42] wgrant, perfect. (For the purpose I'm thinking of, at least. Not necessarily Colin's.) [15:42] +copy-packages would be improved by having to toggle in the PPA name in octal [15:42] Indeed [15:42] It's probably from back in the good old days when PPA names were immutable. [15:43] (Also, IIRC it defaults to the copy target being the same PPA and same series.) [15:43] (although that still would have run into the duplicate person display name issue) === salgado is now known as salgado-lunch [15:45] sinzui: i can. [15:45] thank you. [15:46] I am switching dns servers at this moment so I was not certain I missed your reply [15:50] sinzui: should it be marked qa-bad bad-commit &c? [15:51] (the bug on said rollback branch) [15:51] jcsackett: yes, for our purposes [15:51] sinzui: ok. [15:51] i'll tag it as well. [15:52] jcsackett: we cannot prove it is safe to release so we do not want it to go out with the sharing code [15:53] dig. [15:57] sinzui: two more things. do you want to look at the MP, or are you cool with rs=me, and do we know if a rollback of this branch is safe to bzr lp-land or does it need to play on ec2? [15:58] Should be safe to lp-land. [15:58] cjwatson: thank you [15:58] Nothing else since then has touched related code. [15:59] jcsackett: use rs=you since cjwatson is the authority in this case [15:59] cjwatson: thanks. [15:59] sinzui: got it. [16:08] sinzui: it's off. [16:09] thank you very much [17:13] jcsackett: cjwatson: buildbot now sees my restore of rev 15823. I think everything is now good. I have some hope another branch will land in the next few hours to justify the build we have queued in build bot. === salgado-lunch is now known as salgado [17:15] Phew. === almaisan-away is now known as al-maisan [19:12] deryck: I'm the OCR. Do you have the time to review https://code.launchpad.net/~abentley/launchpad/upgrade-badly-stacked/+merge/120854 ? [19:13] abentley, sure [19:23] abentley, heh. "I have a LOC credit of 1860." Now you're just showing off there. ;) [19:23] deryck: :-) [19:50] I believe wgrant is -26,000 LoC. I an -10,000 LoC since we started Disclosure [19:50] PS. The squad is as whole is +10,000 because of the new features [19:52] is there a black market in LoC? [19:53] if not, should I file a bug about implementing one in LP itself? [19:53] elmo: I can hook you up. There's some code that I know is going obsolete soon :-) === al-maisan is now known as almaisan-away [20:13] It's like a bunch of MMO gold farmers up in here. [20:13] abentley, r=me, looks good. [20:13] deryck: thanks. [20:18] fab, I was going to as if that branch needed a review === jelmer is now known as Guest73241 === jpds is now known as Guest81255 === salgado is now known as salgado-brb === salgado-brb is now known as salgado === Guest73241 is now known as jelmer [22:57] wgrant: http://pastebin.ubuntu.com/1161748/ [23:46] wgrant, wallyworld: What kind of sanity check does reconcile_access_for_artifact need? [23:47] StevenK: when adding a new bug target, the target project may not have the relevant access policy defined [23:48] so a check needs to be done that pillar count == policy count (in simplistc terms) [23:48] wgrant says we should just raise an exception in that case [23:48] >= surely [23:48] yes [23:49] other way round perhaps [23:49] Ah, no, we need to make sure that each pillar has an AP. [23:49] And if no, croak [23:49] each pillar needs to have an access policy for the relevant bug information type [23:55] wallyworld, StevenK: I think an exact match is best [23:56] reason being? [23:57] There should be exactly one for each [23:57] If there isn't, something is wrong [23:57] If something's wrong, better to crash than corrupt [23:57] (eg. a private junk branch is being transferred to a public team somehow) [23:58] ok, be interesting to see how many times we need to raise an exception in practice