[01:05] <StevenK> wgrant: Thoughts about http://pastebin.ubuntu.com/5632908/ for bug 888049 ?
[01:05] <_mup_> Bug #888049: garbo-frequently logged an oops <oops> <Launchpad itself:Triaged> < https://launchpad.net/bugs/888049 >
[01:06] <StevenK> If you say it's good, I'm going to self-review the branch and land, but want your thoughts first.
[01:16] <wgrant> StevenK: It's not ideal, but it's probably OK
[01:16] <wgrant> We might want some sort of threshold
[01:16] <wgrant> But this is probably OK for now
[01:18] <StevenK> wgrant: Threshold? For what?
[01:18] <StevenK> That warning only fires if a tuner is getting killed because the script is.
[01:19] <wgrant> StevenK: Sure
[01:19] <wgrant> StevenK: But we probably sort of want to warn if it legitimately doesn't finish
[01:19] <wgrant> But not if it was blocked all the time
[01:19] <wgrant> Anyway, this seems OK
[01:19] <wgrant> for now
[01:34] <StevenK> Oh, wait a moment.
[01:35] <StevenK> The bug is about the looptuner warning, not the script warning.
[05:29] <StevenK> wgrant: https://code.launchpad.net/~stevenk/launchpad/garbo-abort-info/+merge/154592
[05:32] <wgrant> StevenK: Can you adjust the logger's level so the tests can test what they're testing?
[05:42] <StevenK> wgrant: It's a nice plan, but doesn't love me much.
[05:51] <StevenK> wgrant: http://pastebin.ubuntu.com/5633274/ fails
[05:51] <StevenK> wgrant: The INFO lines aren't printed, but everything else for those blocks is
[05:56] <wgrant> StevenK: What if you pass a custom logger into the tuner?
[05:57] <StevenK> If I create a BufferLogger, set its level to INFO and then make use of loop.log.use(), it also doesn't print INFO line
[05:59] <wgrant> StevenK: Even if you print the content of the buffer?
[06:09] <StevenK> wgrant: http://pastebin.ubuntu.com/5633308/
[06:12] <wgrant> StevenK: I would suggest overriding the tuner's log argument
[06:15] <wgrant> """        Why a custom BatchNavigator is required is a great mystery and
[06:15] <wgrant>         should be documented here.
[06:15] <wgrant>         """
[06:26] <StevenK> wgrant: http://pastebin.ubuntu.com/5633323/
[06:26] <StevenK> And oh dear
[06:28] <wgrant> StevenK: Have you looked at the script logging infrastructure to see how it does the level override?
[06:29] <wgrant> It probably has some custom filter preventing setLevel from working
[06:30] <StevenK> wgrant: I had a little bit of a dig, but not much.
[06:30] <StevenK> The logger stuff is ... odd
[08:45] <czajkowski> aloha
[08:57] <jam> frankban: looking here: https://launchpad.net/~gophers/+related-projects there is "golxc" is that something that should stay in ~gophers or move to ~juju?
[09:12] <frankban> jam: I don't know, I suggest to ask on #juju-dev
[09:12] <jam> sorry frankban, I was in the wrong window. Thanks for understanding
[09:13] <jam> (I was looking for Frank in #juju-dev, but didn't notice the irc window)
[09:13] <frankban> :-)
[21:23] <marcoceppi> Hi all, I'm having a weird issue using launchpadlib on Windows
[21:24] <marcoceppi> Whenever I run the script using launchpadlib I get this SSL error http://paste.ubuntu.com/5635134/
[21:24] <marcoceppi> Happens on Windows Server 2012 on EC2 and a local Windows 7 machine
[21:30] <wgrant> marcoceppi: You probably need to override the CA certificate list path
[21:31] <marcoceppi> wgrant sounds straight forward, would this need to happen in the code or on the machine?
[21:33] <wgrant> marcoceppi: SYSTEM_CA_CERTS in lazr.restfulclient._browser
[21:34] <marcoceppi> wgrant I replaced C:\Python27\Lib\site-packages\httplib2\cacerts.txt with latest from upstream without success.
[21:35] <wgrant> marcoceppi: You'll need to change the path in the module that I referenced.
[21:35] <marcoceppi> wgrant: thanks
[21:39] <marcoceppi> wgrant: so, bascially because SYSTEM_CA_CERTS is hardcoded n lazr.resfulclient._browser it'll just never be portable beyond Ubuntu?
[21:42] <wgrant> marcoceppi: There's no standard way to find a list of CA certifactes, unfortunately. A list of fallbacks could be added if someone interested did it, but that's all we have today.
[22:10] <marcoceppi> wgrant I've got a small patch that'll at least make lazr.resfulclient work on non-debian/ubuntu platforms while maintaining proper ubuntu support https://code.launchpad.net/~marcoceppi/lazr.restfulclient/system-ssl-certs/+merge/154821
[22:14] <wgrant> erm
[22:14] <wgrant> When did httplib2 become not totally insane...
[22:14] <wgrant> Huh, that actually works
[22:15] <wgrant> I guess someone must have finally convinced Python people that making it extremely awkward to use HTTPS was actually a stupid idea :)
[22:15] <wgrant> httplib2.CA_CERTS has only existed for about a year, it seems
[22:15] <wgrant> marcoceppi: It might be worth having it the other way around
[22:16] <marcoceppi> wgrant FWIW, CA_CERTS works on Windows, though I haven't tested that path directly on Ubuntu
[22:16] <wgrant> See if httplib2.CA_CERTS is importable, and if it is use that. Otherwise catch the ImportError and use the hardcoded path
[22:16] <marcoceppi> wgrant, ah that makes more sense. I'll update my merge request
[22:16] <wgrant> marcoceppi: The path is correct on Ubuntu too; the reason we didn't use it is that it didn't exist
[22:16] <wgrant> Everyone hardcoded their own paths because httplib2 was stupid :/
[22:16] <wgrant> But they finally saw sense
[22:17] <marcoceppi> glad they wised upppppp
[22:17] <marcoceppi> wow, keyboard lag
[22:17] <wgrant> Heh
[22:19] <marcoceppi> wgrant merge updated
[22:24] <wgrant> marcoceppi: Hm, actually, look at the NEWS entry in http://bazaar.launchpad.net/+branch/lazr.restfulclient/revision/134
[22:24] <wgrant> marcoceppi: Maybe httplib2's been fixed, if it works on Windows for you?
[22:25] <marcoceppi> It does work on Windows for me now, however I realize I don't have a standard cacerts.txt file. I downloaded a new one from upstream. Let me see what was bundled
[22:25] <marcoceppi> *let me try what came bundled
[22:26] <marcoceppi> Nope, the one that came with httplib2 that was downloaded earlier today works fine
[22:27] <marcoceppi> I'm able to get responses from lp, etc. So rev 134 just needs to be reversed?
[22:27] <wgrant> Which version of httplib2 are you using?
[22:28] <wgrant> For safety we might want to go with what you originally had, since old Ubuntu releases might have a bad httplib2.
[22:28] <marcoceppi> 0.8-py2.7
[22:28] <wgrant> Ah
[22:28] <wgrant> We have 0.7.2ish
[22:28] <marcoceppi> Ah, so the original commit would have been better
[22:29] <marcoceppi> Actually, shouldn't the import of CA_CERTS still throw an ImportError if it's not in 7.2ish
[22:29] <wgrant> Er yeah
[22:29] <wgrant> With the ImportError catching
[22:29] <wgrant> Indeed
[22:30]  * marcoceppi double checks
[22:30] <wgrant> So if the existing hardcoded path exists, use that. Otherwise try to import httplib2.CA_CERTS and use that.
[22:30] <wgrant> So basically what you had before, except with the CA_CERTS import guarded
[22:31] <marcoceppi> wgrant cool, I'll update the merge again
[22:31] <wgrant> Thanks.
[22:32] <marcoceppi> So, for the catch, shouldn't it actually raise an error. Because there's no CA_CERTS, and no hardcoded path available
[22:33] <marcoceppi> I feel like it'd be better to throw a more sane error there than just having the current error which is really misleading
[22:34] <wgrant> Possibly. Even the ImportError would probably be more sensible than what's there now
[22:34] <wgrant> So maybe don't catch the exception, but only try the import if the hardcoded path doesn't exist?
[22:34] <marcoceppi> wgrant: ack
[22:34] <StevenK> Which then results in people having to know why _browser is throwing an ImportError
[22:35] <wgrant> StevenK: It's going to be fairly clear
[22:35] <wgrant> Because the line will be 'from httplib2 import CA_CERTS' :)
[22:35] <czajkowski> StevenK: morning
[22:37] <StevenK> czajkowski: HAI
[22:41] <marcoceppi> wgrant: rebased and resubmitted the merge request https://code.launchpad.net/~marcoceppi/lazr.restfulclient/system-ssl-certs/+merge/154826
[22:44] <marcoceppi> Thanks for the help!
[22:44] <wgrant> marcoceppi: looks good, but I have suggested an optional change in the MP
[22:44] <wgrant> I see you've already signed the contributor agreement, so I can land this right away if you want
[22:44] <marcoceppi> wgrant: I'll just implement your change, and you can feel free to land
[22:44] <wgrant> Thanks, will do
[22:46] <marcoceppi> wgrant: awesome, this will help move things along
[23:08] <wgrant> marcoceppi: Actually, I've just had yet another thought
[23:08] <wgrant> It presumably defaults to CA_CERTS internally...
[23:08] <marcoceppi> presumably
[23:09] <wgrant> But it wouldn't really be any cleaner to avoid overriding the default, so I guess this is fine
[23:09] <wgrant> Nevermind :)
[23:10] <marcoceppi> Hopefully when all ubuntu distros are at >httplib2-0.8 this whole workaround can just be dropped. At least until httplib2 does something silly again
[23:11] <wgrant> All supported releases should have httplib2.CA_CERTS already, I think, but I don't want to rely on it just yet.
[23:12] <wgrant> (0.7.2 was backported to -security on the basis that the old one was basically impossible to secure)
[23:12]  * marcoceppi nods
[23:43] <StevenK> wgrant: What was your objection about http://pastebin.ubuntu.com/5635613/ ?
[23:55] <wgrant> StevenK: I don't think I saw that.
[23:55] <wgrant> 'cause that's basically ideal :)