[01:03] StevenK: So, how far through the lazr.restful source have you been? [01:03] wgrant: I've stared at exported() for a while and it still makes little sense [01:04] StevenK: It adds annotations. But you need to look at what interprets those annotations === lifeless_ is now known as lifeless [01:12] wgrant: What does the interpreting? [01:13] StevenK: Yes [01:25] wgrant: While talking with lifeless I've been combing lazr.restful and still coming up with no dice [01:26] I'll look shortly [01:27] * StevenK finds another bug [01:31] Dear AAPT, please DIAF [01:31] Maybe my IPv4 connectivity out of Australia will work now [01:42] wgrant: Are you going to request a deploy of oops-tools? [01:49] StevenK: I need to talk to lifeless about how we deploy/release/whatever oops* [01:49] hi [01:49] you just ask bebops to do it [01:49] there isn't a wiki page yet. [01:49] Perhaps there should be [01:50] lifeless: We need to release to PyPI first, or do they deploy from branches? [01:57] lifeless: ^^ [01:57] they deploy from branches, but please release to PyPI [01:58] we should split oops-tools like we did auditor [01:58] would make that better [01:58] Split? [01:58] Also, I think only you have PyPI privileges [01:59] I'm not even sure I have an account [02:05] wgrant: also I have some oops-tools stuff pending [02:05] adding query capturing [02:05] wgrant: split into django site and django app, and make the site internal only [02:07] lifeless: Ah, right [02:07] BLEH, bug 1020439 leads straight back into lazr.restful [02:07] <_mup_> Bug #1020439: AttributeError: 'NoneType' object has no attribute 'items' in +preview-diff < https://launchpad.net/bugs/1020439 > [02:07] james_w: hi [02:08] wgrant: get thyself a pypi user id [02:08] wgrant: I've added stevenk for now [02:09] wgrant: pypi thinks there is a wgrant btw, if its not you say so asap :) [02:10] lifeless: Ah, that is me indeed. [02:11] I see that I own oops-tools now [02:11] Thanks [02:11] give me a few hours to kick the tires on this query capturing [02:11] Sure [02:11] I'll be very happy when you get that done :) [02:12] lifeless: I also need oops_datedir_repo [02:12] wgrant: So, your choice, help with security_contact, or peer at OOPS-e8d958307abcc91627fcb35e08fff15e :-) [02:12] StevenK: My only two options involve obscure lazr.restful? You are a cruel man. [02:13] StevenK: What have you uncovered so far? [02:13] wgrant: whats pending there ? [02:14] lifeless: The same fix [02:14] wgrant: For either? That I don't understand lazr.restful much :-( [02:15] wgrant: gimme a hint [02:16] wgrant: 0.0.18 is published on pypi already, and there is nothing else committed to trunk for datedir-repo [02:16] btw, I realised you can meaningfully salt anonymiseip [02:17] Its now in my todo list to add a salt option [02:17] Ah, good :) [02:17] Um, I approved the datedir-repo MP, maybe tarmac hasn't noticed [02:17] Or there is no tarmac [02:17] there is no tarmac [02:17] python-oops-tools has one, so I thought there might be for datedir-repo [02:18] the README notes this, the trunk branch URL tells you as well [02:18] the owner of the branch is launchpad-pqm for tarmaced things [02:18] Yes, but I didn't bother to check since oops-tools worked :) [02:18] hah [02:18] ok [02:18] I see [02:18] this is the prune SNAFU [02:19] Right [02:19] so, do the dance to do a 0.0.19 release of datedir-repo [02:19] no need to wait for me, and look at the commits in the log for 0.0.18 and the rev after it to see how its done. [02:19] For bonus points, write a damn tool to do releases. [02:20] Heh [02:21] then you'll want to do another patch to oops-tools [02:21] To upgrade versions.cfg, right [02:21] to bump the datedir-repo version [02:28] wgrant: https://code.launchpad.net/~lifeless/python-oops-tools/timeline/+merge/125086 [02:29] wgrant: out for abit [02:29] lifeless: k [02:48] lifeless: Your MP seems to lack a dep on timeline_django in setup.py [03:05] hi lifeless [03:41] StevenK: So, those two lazr.restful issues [03:41] StevenK: I know not much more about lazr.restful than you do [03:41] wgrant: Heh, handy [03:42] * StevenK has moved onto bug 827244 for the moment [03:42] <_mup_> Bug #827244: bug_filters related to teams are said "mutable" when they're not < https://launchpad.net/bugs/827244 > [03:42] wgrant: I'd go back to using as_of='1.0' except I'm stubborn and don't want security_contact in devel [03:42] wgrant: django_timeline you mean ? [03:43] lifeless: No, timeline_django [03:43] oh erm [03:44] * wallyworld_ had fun watching bigjools and lifeless have a public belching competition at lunch. real classy [03:44] wgrant: thanks [03:44] wgrant: we thought so [03:45] bah [03:45] wallyworld_: ^ [03:45] Hahaha [03:47] wgrant: Are you handing QA for r15973? [03:47] StevenK: You're 38 seconds late [03:48] Oh, sorry, I was looking at doing the QA for r15972 [03:48] A good idea [03:51] StevenK: Hm [03:51] StevenK: For security_contact, did you try adding a version annotation of ('devel', dict(exported=False))? [03:51] With QA done I'm poking around the source [03:51] And that looks plausible [03:51] wgrant: so, I think we'll definitely find it captures stuff it shouldn't from the openid auth module for django [03:52] wgrant: i'm not sure where to start for neutering that [03:52] lifeless: Indeed. [03:52] wgrant: I did not [03:52] lifeless: Personally I think capturing parameters is generally pretty insane. [03:52] wgrant: yes, because no one ever does manual SQL. [03:52] wgrant: So you think "('devel', dict(exported=False)), as_of='1.0'" ? [03:53] StevenK: I forget if you can replace as_of itself with a version annotation, but yes, that should work [03:53] lifeless: If they do that in Django apps then I will be sad. [03:56] wgrant: Ah ha! [03:56] StevenK: Are you still QAing that essential sub thing? [03:56] Looks pretty trivial [03:56] then we can deploy [03:57] And private teams can be less slow [03:57] wgrant: Trying to work out how, sadly [03:57] I can still tick essential for ubuntu blueprints [03:57] wgrant: so, should we land it, find out the issue, file a bug and comment out the hook ? [03:57] So I'm trying to work out what the catch is [03:58] lifeless: I'd comment out the middleware addit [03:58] ion [03:58] wgrant: and then have ops uncomment it ? [03:59] I don't know :) [03:59] wgrant: so the middle ware isn't what captures the data [03:59] Safety on capturing SQL statements is an unresolved issue. [03:59] True [03:59] wgrant: its the timeline hooks that does it. [03:59] which is what I was suggesting commenting out to disable [04:00] timeline_django.filters.install_hooks? [04:00] timeline_oops [04:00] wgrant: Right, I don't get it. [04:00] the one you listed does sanitisation for the stock django bits. [04:00] StevenK: The subscriber can no longer tweak the flag [04:00] StevenK: Or isn't meant to be able to [04:01] wgrant: I can set it when I subscribe [04:01] StevenK: what about afterwards? [04:01] james_w: please be doing a release of timeline_django :) [04:01] lifeless: Which bit is it that logs SQL to the timeline, then? [04:01] wgrant: If I hit Update subscription I get 'Not allowed here' [04:01] timeline_django_setup? [04:01] Which is just awesome [04:02] StevenK: Right, so that's hideous but the logical effect of the revision [04:02] StevenK: Even if you don't change the flag? [04:02] wgrant: the middleware [04:02] The revision's commit message also doesn't match its content [04:02] wgrant: I can't even see the page to change the flag [04:02] 13:59:21 < lifeless> wgrant: so the middle ware isn't what captures the data [04:02] wgrant: I'm saying disable the entire timeline when there is an issue [04:02] wgrant: the timeline isn't in the oops by default [04:03] wgrant: I also can't unsubscribe myself from the blueprint [04:03] wgrant: middle ware -> hooks into django and adds to timeline; timeline hooks hook timeline into the oops [04:03] Oh, disable the timeline from being included in the OOPS, I see [04:03] StevenK: rofl [04:03] StevenK: Revert revert revert [04:03] And I can't deploy my private team fix :( [04:03] lifeless: Right, makes sense now [04:04] wgrant: Sorry [04:04] wgrant: so, plush one on my mp ? [04:04] lifeless: Do we want an I_ACKNOWLEDGE_THAT_I_AM_TERRIBLY_MISGUIDED flag in settings.py to enable it, perhaps? [04:04] lifeless: Don't we want it disabled in trunk? [04:04] That MP still has it enabled. [04:07] lifeless, the only change is your debugging aid? [04:07] james_w: yes, which would be very useful for oops.c.c [04:08] lifeless, you can't debug the issue locally? [04:08] wgrant: thats right, its safe enough for folk not running custom auth crazy [04:08] james_w: nope, can't. [04:08] lifeless, you had a reproducer though? [04:08] james_w: can't reproduce now [04:08] james_w: driving me batty [04:08] ok [04:08] not tonight [04:08] maybe not tomorrow [04:08] lifeless: Is it? [04:08] lifeless: But our auth crazy is at the Apache level anyway [04:09] wgrant: is it ? I don't think it is. [04:09] Oh, you mean the /admin thing, yeah [04:09] Maybe [04:09] james_w: cool, let someone here know - purple or me - and we'll bump the dep in oops tools at that point. [04:11] lifeless, poke me tomorrow or the next day if I haven't [04:11] we're in a bit of a crunch [04:11] ok [04:12] lifeless: btw, this version-locked madness is madness [04:12] wgrant: Revert landed as r15980 [04:13] StevenK: Thanks [04:13] wgrant: Shall we deploy up to r15971? [04:13] We have little choice [04:13] StevenK: Did the security_contact thing work? [04:14] wgrant: It worked so well I put it up for review. https://code.launchpad.net/~stevenk/launchpad/export-security_contact-for-1.0/+merge/125092 [04:14] I got an email right as you said that [04:14] lifeless: I have a tested datedir-repo sdist -- can I have upload privs now? [04:15] StevenK: This doesn't sound like it will fix the proejct creation script [04:15] StevenK: As it's RO [04:15] Right? [04:16] wgrant: done [04:16] I have no idea about the project creation script, the bug only talked about exporting it [04:17] lifeless: Thanks [04:17] StevenK: The bug description mentions it [04:20] wgrant: so.. +1? [04:21] wgrant: Sure, but it doesn't say that it has to be settable [04:21] lifeless: k [04:21] done [04:21] StevenK: A project creation script presumably wants to set it [04:22] lifeless: How do I add a new release on PyPI? [04:22] wgrant: Let's leave it for now, I'll bring it up on the call [04:22] I try to add a new file, but it lists the 0.0.18 tar.gz, so I'm not sure if it's adding a new release or not [04:22] I'm also slightly nauseous from uploading to a totally insecure distribution mechanism [04:24] wgrant: so, you make the release by doing [04:24] setup.py sdist upload -s [04:24] wgrant: gpg signed, ~= totally insecure [04:24] OK, all the clients are totally insecure :) [04:24] Now how do I auth... [04:26] cat .pypirc [04:26] [server-login] [04:26] username:lifeless [04:26] password: [04:27] Oh cool [04:27] It defaults to basic auth over unencrypted HTTP [04:27] yup [04:28] And I bet if asked to do HTTPS it doesn't actually check the cert [04:28] patches... [04:29] Yeah, no cert verification [04:31] Ah [04:31] Not that it really matters [04:31] Since you can change your password through the web UI with just a cookie [04:32] Yeah, security is for other frameworks [04:33] I think it's uploaded [04:33] I might need to go outside to breathe before I refresh the page, though [04:34] The hash even matches [04:35] wgrant: or just take a drink every time there's some security thing that annoys you. pretty soon you won't care at all :) [04:35] Hahaa [04:35] They're not security things that annoy me, they're security things that are objectively wrong :) [04:35] ajmitch: And after using LP for 15 minutes, wgrant gets admitted to hospital for alcohol poisioning [04:36] LP is merely less than ideal [04:36] Other things are trainwrecks :) [04:39] Oh damn, I looked at a branch page wrong during initial scan [04:44] bigjools: http://www.xtranormal.com/watch/7023615/watch_movie <- kick this one off next. [04:48] wallyworld_, wgrant: https://code.launchpad.net/~stevenk/launchpad/mute-when-not-allowed/+merge/125094 [04:52] wgrant: I think bug 869351 and bug 1036882 are dupes. Possibly bug 816617 too [04:52] <_mup_> Bug #869351: UnknownRecipientError on +editstatus page when setting assignee, milestone, status, and commenting < https://launchpad.net/bugs/869351 > [04:52] <_mup_> Bug #1036882: UnknownRecipientError OOPS updating series specific tasks < https://launchpad.net/bugs/1036882 > [04:52] <_mup_> Bug #816617: UnknownRecipientError raised sending mail notification to bug subscriber < https://launchpad.net/bugs/816617 > [04:56] StevenK: I believe they are probably dupes, yeah, but I didn't get around to digging farn enough into them on Monday to tell for sure [04:57] Hmk [04:57] I thought I duped 1036882 against 869351 on monday [04:57] But apparently not [04:58] Ah, and the OOPSes from the older two are gone [04:58] Yeah [04:59] Of course [04:59] I think my diagnosis in 1036882 can reasonably apply to the other two [04:59] * wgrant dupes [04:59] I was about to, but sure [05:00] Ah right [05:00] wgrant: Your diagnosis of "I'm not sure why there was no reason." ? :-) [05:00] I duped 925791 [05:00] So there were 3 duplicates [05:00] StevenK: can you put the team name in the error message and maybe word it like the message just below in the code? [05:01] Also, does LP prefer "can not" or "cannot"? [05:01] I suspect the latter [05:02] +1 [05:02] wgrant@lamuella:~/launchpad/lp-branches/devel$ bzr grep '\bcan not\b' | wc -l [05:02] 77 [05:02] I think we have a clear winner [05:02] wgrant@lamuella:~/launchpad/lp-branches/devel$ bzr grep '\bcannot\b' | wc -l [05:02] 1298 [05:03] I wish bzr grep supported -c [05:03] Don't make me go full UNIX on you [05:07] wallyworld_: http://pastebin.ubuntu.com/1214202/ [05:07] wgrant: so, you're doing the oops-tools repo dep upgrade? [05:07] wgrant: or do you want me to ? [05:08] StevenK: thanks, s/muted, team/muted because team [05:09] wallyworld_: http://pastebin.ubuntu.com/1214207/ [05:09] lifeless: I guess some of the other deps need upgrading too [05:09] r=me, thanks [05:10] lifeless: Since datedir-repo is at 0.0.15... [05:10] maybe [05:10] you'll get conflicts if they are incompatible [05:10] or something [05:38] jam: whats your pypi id ? [05:39] lifeless: jameinel [05:40] jam: I'd like to apply http://paste.ubuntu.com/1214228/ to bzr-pqm and do a release on pypi. [05:41] lifeless: do we actually require launchpadlib? I thought it was a nicety, but for people who might use it on Windows it brings in a huge dependency set. [05:41] Otherwise +1 [05:41] jam: it selftest fails w/out it. [05:41] jam: I don't recall if it fails to load at all w/out it. [05:41] I know I used 'bzr pqm-submit' without it for a long time [05:42] wgrant: having some problems with “bzr bd” on maas… are you familiar with it? [05:42] but certainly the lp- versions of the pqm commands probably require it [05:42] (lpland, etc) [05:42] jtv: Only vaguely. [05:42] jtv: I gather you have some people who know bzr pretty well now, so they probably know it better than I [05:42] But I can help if you need [05:42] anyway, I won't block it, and people using bzr-pqm likely have launchpadlib around, there aren't many outside of lp people using pqm anyway. [05:42] wgrant: I also do not get why ... (re: bug 1036882) [05:42] <_mup_> Bug #1036882: UnknownRecipientError OOPS updating series specific tasks < https://launchpad.net/bugs/1036882 > [05:43] lifeless: are you asking me to apply the patches, or you need pypi access, or just a review? [05:43] StevenK: Have you been able to reproduce it locally? [05:43] jam: I could move it to extras, but I'm not sure there is a good way to say 'bring it in for this command' other than splitting the package. [05:43] wgrant: No, I've been reading the code [05:43] jam: eyeball review, since I plan to commit, push, sdist upload register etc all at aonce. [05:43] wgrant: fwiw Julian is stumped. :) A patch fails to apply, and I get the impression it should. I suspect the fact that one of the paths has a double “.orig” in it. [05:43] jam: so, say '+1' now,a nd then you can forget about it :) [05:43] lifeless: so I'm hesitant about launchpadlib, but I understand and I can live with it for your +1 [05:43] wgrant: All of the subscriber (as in zope subscriber) and the recipient fetching code [05:44] jam: they can't currently pip install it anyhow, as its not on pypi at all :) [05:44] StevenK: Have you set up the scenario described in the bug locally? [05:44] wgrant: the “---” file is maas-0.1+bzr702+dfsg.orig.orig/contrib/maas_local_settings.py [05:44] jam: which this will at least solve; can fine tune later :> [05:44] jam: thanks [05:44] wgrant: No, that's my next step [05:46] wgrant: So, create a bugtask and milestone, create a structural subscriber for the milestone and +editstatus? [05:46] StevenK: Something like that. Possibly with a target change, or assignee change, or something like that [05:46] The full circumstances should be described in the bug [05:46] I was going to add those in, yes. [05:47] jam: and there - http://pypi.python.org/pypi/bzr-pqm [05:47] done [05:53] lifeless: looks good [05:53] thanks for doing the work [05:53] this is part of 'get rid of sourcecode' ? [05:54] launchpad guys, I get initial emails when someone submits a MP to lp:maas, however, I do not get comments on all MPs. I think I subscribed to the branch, is there a config I'm missing to get the followup emails? [05:54] jam: its part 'get rid of inventory' [05:54] jam: and part 'make it easier to get rid of sourcecode' [05:54] lifeless: 'inventory' being things that aren't properly released yet? [05:55] jam: we either need to switch to e.g. pip (something that doesn't use separate egg dirs per thing), or we need to teach bzrlib how to load plugins from eggs. [05:55] jam: yup [05:55] jam: 'fix committed' status [05:55] jam: I mailed the list about the issue [05:56] lifeless: 'pip install -Z' doesn't build a .zip/.egg file, I believe. Can't you set the 'not zip safe' flag somewhere in setup.py? [05:56] jam: EPARSE [05:57] jam: pip doesn't use egg directories at all, it installs into site-packages. [05:57] lifeless: sorry, "easy_install -Z" then. The thing that creates eggs can do it by just updating .pth and then putting the files expanded on disk. [05:57] ok, gotta run, taxi here. [05:57] I used to have to do that for paramiko [05:57] because otherwise py2exe couldn't bundle it. [05:58] lifeless: catch you later. [06:16] wgrant: this “bzr bd” problem with maas is really bugging us. A patch fails to apply to a missing file, except the missing file seems to be in the branch as before. [06:17] An invocation of “dh get-orig-source” seems to be part of the problem. There seems to be no such command in dh. [06:18] I strongly doubt it's dh [06:18] dh is a debhelper wrapper [06:18] Are you perhaps trying to use get-orig-source in a package that doesn't support get-orig-source? [06:19] bzr bd is after my time, anyway [06:20] How would I tell whether the package supported it? There is a target of that name in debian/rules though. [06:20] That is roughly the definition of supporting it. [06:20] wgrant: http://pastebin.ubuntu.com/1214252/ and no failure :-( [06:20] Not sure why it would be trying to call into dh [06:20] the problem is when quilt is applying the patches [06:20] beyind that I dunno [06:21] It doesn't find the file it's supposed to be patching. To be honest, I can't find it either. It's in the .orig tarball, but I don't see it unpacked anywhere. [06:23] Is it deleted by the .diff.gz applying? [06:24] maas hopefully doesn't have a diff.gz [06:24] Or an .orig then [06:25] It's just .tar.gz for native [06:26] How would I check these things? I see no files with any of those names in the packaging branch, fwiw. [06:26] StevenK: I was alluding to the fact it is probably 3.0 (quilt), which cannot delete files from the orig.tar.gz [06:26] Ah, I've been ignoring 3.0 (quilt) for being a crime against humanity [06:27] wgrant: Ah ha, my save against BugTask:+editstatus doesn't hit the zope subscriber stuff [06:29] Well the file that's being patched is still in the .orig.tar.gz. [06:34] wgrant: I guess it's normal that, after the failed build, the build-area directory contains the .orig.tar.gz and the packaging branch, but not the contents of the .orig.tar.gz in unpacked form? [06:34] jtv: I'm really not sure, sorry [06:34] Any idea who might know, that's available at this hour? [06:37] wgrant: Hm, I thought passing form=foo to create_initialized_view() would call the save_action [06:38] StevenK: Perhaps your action name is bad [06:38] StevenK: Does the form appear to submit? [06:38] jtv: Perhaps former bzr people may know [06:38] Ah ha, view.errors [06:39] wgrant: thanks, I think I'll be able to dig some up soon! [06:51] wgrant: The form submits, zope subscriber stuff fires, no error [06:53] StevenK: Have you tried manually setting up the scenario on launchpad.dev? [06:53] You could just have a bad test [06:54] wgrant: http://pastebin.ubuntu.com/1214277/ [07:07] StevenK: I can't always solve all your problems just by looking at a test of a strange view, sadly [07:08] StevenK: Try it in a browser [07:08] wgrant: I am doing so [07:08] Trying to remember how to add a series task [07:09] Target [07:10] Don't I have to nominate? It's been so long since I've had to do stuff like this. [07:10] Not if you're the driver [07:11] Which I'm not [07:11] Let me fix that [07:13] wgrant: No OOPS [07:13] You won't even see the link if you're not one of maintainer/driver/bugsupervisor [07:13] Hmm [07:13] I have a product, which I'm driving, a series, a milestone and a structsub for a random on the milestone [07:15] StevenK: Do you have an appropriate event filter? [07:15] StevenK: The problematic subscription on production was open/close only [07:15] That isn't the default? [07:17] StevenK: Not through the API, probably. [07:17] Or through the method you called in the test [07:18] Right, I'll check the filter [07:21] Hm, how do I even express that as a filter? :-( [07:23] StevenK: It's one of the root settings for the subscription [07:24] StevenK: So it's not shown as a filter in the web UI [07:24]