| lifeless | bac | 00:15 |
|---|---|---|
| lifeless | *back* | 00:15 |
| lifeless | (sorry bac) | 00:15 |
| StevenK | lifeless: Haha, I did wonder if you were at your keys or wanted Brad's attention | 00:16 |
| === lifeless changed the topic of #launchpad-dev to: Performance Tuesday | https://dev.launchpad.net/ | On call reviewer: - | Critical bugs: 240 - 0:[#######=]:256 | ||
| === wgrant changed the topic of #launchpad-dev to: Performance Tuesday | https://dev.launchpad.net/ | On call reviewer: wgrant | Critical bugs: 241 - 0:[#######=]:256 | ||
| wgrant | Anyone around to review https://code.launchpad.net/~wgrant/launchpad/iseriesbugtarget/+merge/69951? | 00:32 |
| wgrant | lifeless: You don't have a good suggestion for a primary bug target? | 01:08 |
| wgrant | ie. product/distro/DSP? | 01:09 |
| lifeless | wgrant: 'thing', 'target', 'context', shrug. | 01:09 |
| lifeless | wgrant: I think we want to remove the series special casing eventually, so I don't really care about the name. | 01:10 |
| wgrant | True. | 01:10 |
| wallyworld | lifeless: following on from last week, could you please look at this mp (and +1 if you are happy). then i will land it for the contributor https://code.launchpad.net/~xaav/loggerhead/export-tarball/+merge/66408 | 01:45 |
| wgrant | * 9967 Exceptions | 01:58 |
| wgrant | Hi codehosting. | 01:58 |
| lifeless | \o/ | 01:58 |
| poolie | hi wallyworld | 02:08 |
| wallyworld | poolie: hi | 02:09 |
| poolie | thanks for pushing it | 02:09 |
| wallyworld | np | 02:09 |
| wallyworld | i got caught up yesterday | 02:10 |
| === jtv-zzz is now known as jtv | ||
| jtv | StevenK, wgrant: I uploaded a package to -partner (on dogfood) and ran publish-ftpmaster. The pool now contains the .dsc and 2×tarball, and the package is in the Sources list, but it's not showing up in the Packages files and I don't see a .deb. Is that bad? | 02:47 |
| StevenK | Has it built? | 02:47 |
| StevenK | And 2*tarball? | 02:48 |
| wgrant | Needs build + potential acceptance | 02:48 |
| jtv | StevenK: orig.tar.gz and debian.tar.gz | 02:48 |
| jtv | How do I find out about build and potential acceptance? | 02:48 |
| StevenK | What does dogfood.l.n/builders say? | 02:49 |
| jtv | 2 builders: ferraz & rubidium, both idle | 02:49 |
| jtv | Haven't found my package ("bye") in the logs yet, but the package I based it on ("hello") failed to build. | 02:50 |
| jtv | chroot problem. | 02:51 |
| jtv | Ah, "bye" built successfully on ferraz. | 02:52 |
| wgrant | It probably ended up in NEW. | 02:52 |
| wgrant | Yep, there it is. | 02:53 |
| wgrant | https://dogfood.launchpad.net/ubuntu/oneiric/+queue | 02:53 |
| jtv | The source package page also fails to show a deb. | 02:53 |
| wgrant | It doesn't really exist yet. | 02:53 |
| jtv | https://dogfood.launchpad.net/ubuntu/+source/bye/1.0 | 02:53 |
| jtv | So… do I need to approve the binary upload like I did the source upload? | 02:54 |
| wgrant | Yes. | 02:54 |
| jtv | On the way. | 02:54 |
| jtv | There we go. | 02:54 |
| jtv | And now… process-accepted? | 02:55 |
| wgrant | publish-ftpmaster should do that, right? | 02:55 |
| jtv | You have me there. Checking. | 02:55 |
| jtv | Yup, it does. | 02:55 |
| jtv | So then I can just re-run publish-ftpmaster? | 02:56 |
| wgrant | Correct. | 02:56 |
| jtv | It's mentioning the package now: "(Arch Specific)" | 02:59 |
| jtv | I don't know who Arch Specific is, but he seems to be doing something to my package. | 02:59 |
| wgrant | "arch specific" means "not architecture independent" | 03:01 |
| jtv | wgrant: to be really, really honest I did sort of gather it would be something like that. :) | 03:05 |
| jtv | Anyway, it's done now. Which seems suspiciously quick. | 03:06 |
| jtv | Then again, there won't have been any changes outside of -partner. | 03:06 |
| wgrant | And partner is both NMAF and small. | 03:06 |
| jtv | NMAF? | 03:08 |
| wgrant | No More Apt-Ftparchive | 03:08 |
| jtv | Ah, you mean it uses Launchpad's indexing code? | 03:09 |
| jtv | Well, all green lights so far. | 03:09 |
| wgrant | Right. | 03:09 |
| jtv | My package is in Packages for i386, and the +source page shows the i386 build. | 03:09 |
| wgrant | Excellent. | 03:09 |
| wgrant | And it didn't delete 600GB of archive? | 03:09 |
| jtv | Errr | 03:10 |
| mtaylor | lifeless: fix all of my problems! | 03:11 |
| mtaylor | :) | 03:11 |
| jtv | wgrant: the ubuntu-archive dir now contains about 15GB, I think. | 03:11 |
| mtaylor | hey all - the new launchpad haproxy stuff is killing me | 03:12 |
| jtv | I may lose power. | 03:12 |
| wgrant | mtaylor: Oh? | 03:12 |
| wgrant | jtv: Sounds good. | 03:12 |
| mtaylor | wgrant: it's causing tarmac to have its connections disconnected | 03:12 |
| wgrant | mtaylor: It maintains persistent connections? | 03:12 |
| mtaylor | wgrant: which is problematic | 03:12 |
| mtaylor | wgrant: it ddoes | 03:12 |
| wgrant | It probably shouldn't. | 03:12 |
| wgrant | It didn't have long-term connections open yesterday, interestingly... | 03:13 |
| mtaylor | wgrant: well, according to abently, bzrlib expects to stay connected for some reason | 03:13 |
| jtv | wgrant: the biggest chunks being in contents-generationold. About 2GB in /srv/launchpad.net/ubuntu-archive/ubuntu/pool. | 03:13 |
| wgrant | mtaylor: I suspect you want to change your plugin to manage transports a little better. | 03:13 |
| wgrant | mtaylor: It can't expect to remain connected for weeks. | 03:13 |
| mtaylor | wgrant: it doesn't | 03:13 |
| jtv | wgrant: surprisingly, almost the same amount of data in ubuntu-partner. | 03:13 |
| mtaylor | wgrant: it just stays connected while it runs tests | 03:14 |
| wgrant | Ah. | 03:14 |
| wgrant | Someone else is keeping connections open for weeks, so I thought you might be too. | 03:14 |
| mtaylor | wgrant: which sometimes is longer than 50s | 03:14 |
| mtaylor | nope. that shouldn't be us | 03:14 |
| wgrant | So, sounds like your plugin needs to be fixed to not do that. | 03:14 |
| wgrant | We can possibly increase the haproxy timeouts for now, if it is causing big problems for you. | 03:15 |
| wgrant | But only very temporarily. | 03:15 |
| mtaylor | wgrant: rockstar told me to come bug you :) --- and it's actually the core tarmac code | 03:15 |
| mtaylor | tarmac holds the connection around the post-merge hook | 03:15 |
| wgrant | Hmm. | 03:15 |
| wgrant | rockstar: Any reason Tarmac can't handle transports a bit more sensibly? | 03:16 |
| mtaylor | wgrant: https://jenkins.openstack.org/view/Nova/job/nova-tarmac/109487/console | 03:19 |
| mtaylor | or more excitingly: https://jenkins.openstack.org/view/Nova/job/nova-tarmac/109467/console | 03:20 |
| wgrant | That's rather unpleasant. | 03:20 |
| lifeless | mtaylor: Use the source monty! | 03:22 |
| mtaylor | lifeless: these are not the droids you are looking for | 03:22 |
| lifeless | mtaylor: wgrant: so, this is a bzrlib bug. | 03:25 |
| wgrant | :( no criticals in that rollout | 03:25 |
| lifeless | not the holding open | 03:25 |
| lifeless | the handling of the disconnects. | 03:25 |
| lifeless | transports are defined stateless. | 03:25 |
| lifeless | so are RPC calls. | 03:25 |
| mtaylor | lifeless: yay! nothing better than a bzrlib bug! | 03:26 |
| lifeless | so the bug is 'when an idle ssh transport is disconnected, bzrlib should not error' | 03:26 |
| lifeless | or | 03:26 |
| lifeless | 'when an idle ssh transport is interrupted, bzrlib errors' | 03:26 |
| lifeless | the haproxy stuff is showing this up very visibly, but the bug pre-existed - bzr didn't seen keepalives, and if ssh wasn't configured to do so it could naturally happen anyhow | 03:27 |
| lifeless | however, lets up it for now to 10 minutes. | 03:27 |
| * mtaylor filing bzr bug | 03:28 | |
| mtaylor | bug 819604 | 03:30 |
| _mup_ | Bug #819604: when an idle ssh transport is interrupted, bzrlib errors <Bazaar:New> < https://launchpad.net/bugs/819604 > | 03:30 |
| mtaylor | lifeless: you gonna be at the next uds? | 03:32 |
| lifeless | no | 03:33 |
| lifeless | new sprog (we hope) | 03:33 |
| mtaylor | what's a sprog? | 03:34 |
| lifeless | child | 03:34 |
| mtaylor | OH! | 03:34 |
| mtaylor | well, in that case, coming to UDS would be the wrong choice | 03:35 |
| mtaylor | (and hopefully congrats) | 03:35 |
| lifeless | yes, we think so | 03:35 |
| mtaylor | LCA then? | 03:35 |
| lifeless | doubtful | 03:35 |
| mtaylor | gah. I'm starting to think you're avoiding hanging out with me :) | 03:35 |
| lifeless | thats it, totally. | 03:35 |
| lifeless | if you wanted to swing by christchurch after UDS we could put you up for a few days | 03:36 |
| lifeless | bah | 03:36 |
| lifeless | LCA | 03:36 |
| mtaylor | ooh. now _that's_ not a bad idea | 03:36 |
| mtaylor | and yeah - I read LCA there | 03:36 |
| mtaylor | I still haven't been to the south island | 03:36 |
| mtaylor | thinking about hanging out with Stewart for a while pre-LCA | 03:36 |
| mtaylor | speaking of - I REALLY need to submit talk proposal... | 03:36 |
| lifeless | UDS +1 I expect I will be at | 03:37 |
| mtaylor | yay! | 03:41 |
| * mtaylor lists UDS's and LCA as the important things he should attend | 03:41 | |
| nigelb | So, recently I used vagrant. | 03:55 |
| nigelb | I'm wondering how easy it would be to create a vagrant file to set up LP environement. | 03:55 |
| jtv | wgrant: I think partner publication works. Next up is expedited -security publishing. For that I need to copy a package into security, right? | 03:56 |
| rockstar | wgrant, the reason is because bzr doesn't handle transports more sensibly? | 04:06 |
| rockstar | I keep hearing launchpad people pushing back on this, and I understand why, but tarmac pushes as much as it can off to bzr. | 04:08 |
| rockstar | I'm unsure why we're timing out so quickly now. | 04:09 |
| * rockstar sees that lifeless has already covered this, and goes to bed | 04:11 | |
| wgrant | jtv: Right. | 04:17 |
| wgrant | jtv: You can copy from a public PPA, if you want. | 04:17 |
| jtv | Thanks. | 04:17 |
| wgrant | Edit before sending? [y/n]: yConnection to bazaar.launchpad.net closed by remote host. | 04:20 |
| wgrant | Heh | 04:20 |
| StevenK | wgrant: Ouch | 04:21 |
| lifeless | StevenK: hi | 04:48 |
| lifeless | StevenK: https://code.launchpad.net/~stevenk/launchpad/pillarname-vocab/+merge/70091 | 04:48 |
| lifeless | StevenK: vocabs depend on stable ordering to paginate | 04:48 |
| StevenK | It is stable? | 04:50 |
| lifeless | ho ? | 04:50 |
| lifeless | how? | 04:50 |
| StevenK | It appeared to be when I tested it on DF | 04:51 |
| lifeless | its not stable with an order by rank, when two rows can have the same rank | 04:51 |
| lifeless | you either need a more granular rank, or a disambiguator | 04:52 |
| StevenK | Right, then I want rank first and name second | 04:52 |
| StevenK | orderBy='rank- name' ? | 04:52 |
| lifeless | for storm? order_by=(SQL('rank'), PillarName.name) | 04:53 |
| wgrant | Desc(SQL('rank')), PillarName.name | 04:53 |
| lifeless | only if you want it to work | 04:53 |
| StevenK | lifeless: It will fail to land anyway, so I'll fix it once I'm sure no other tests have failed | 05:03 |
| lifeless | cool | 05:04 |
| lifeless | sorry for having to raise this | 05:04 |
| StevenK | I should have thought of it, by rights | 05:05 |
| lifeless | :) | 05:06 |
| LPCIBot | Project devel build #937: STILL FAILING in 5 hr 40 min: https://lpci.wedontsleep.org/job/devel/937/ | 05:27 |
| lifeless | wgrant: fun stuff for oneiric : https://bugs.launchpad.net/ubuntu/+source/lxc/+bug/819621 | 06:07 |
| _mup_ | Bug #819621: lucid container start failure after calling lxc-stop (fails across reboots) <lxc (Ubuntu):New> < https://launchpad.net/bugs/819621 > | 06:07 |
| wgrant | lifeless: I noticed that too, but decided it was probably the avahi or similar that I had installed. | 06:10 |
| wgrant | Upsetting to hear that it happens even when the container is not terribly mangled. | 06:10 |
| wgrant | (avahi sets odd rlimits for itself, which mean more than one instance can't run on a single kernel without a bit of work) | 06:10 |
| wgrant | nproc == 2, what could possibly go wrong. | 06:11 |
| lifeless | also having plymouth in the container is just bong | 06:11 |
| wgrant | Mmm. | 06:12 |
| wgrant | Not necessarily. | 06:12 |
| wgrant | It's not just a splash these days. | 06:12 |
| wgrant | It's the way upstart communicates with the outside world. | 06:12 |
| StevenK | jtv: Are you still playing in dogfood's wading pool? | 06:30 |
| StevenK | Bloody *hell* (Error ID: OOPS-2040DOGFOOD18397) | 07:03 |
| StevenK | How has DF generated 18,000 OOPSes? | 07:03 |
| wgrant | Probably the a-f warnings. | 07:04 |
| StevenK | And the publisher is running, so the webapp is being horridly slow. | 07:06 |
| wgrant | :D | 07:07 |
| StevenK | It can't even load bugs.dogfood.l.n | 07:08 |
| wgrant | wallyworld: Hmm, so the question subscription stuff is based on the *old* Bugs stuff? | 07:17 |
| wallyworld | no, how did you think that? | 07:17 |
| wgrant | "Direct subscribers" and "Also notified" | 07:17 |
| wgrant | Old, but more sensible, terms. | 07:18 |
| wallyworld | that's how questions subscriptions have always worked. i just made it use ajax | 07:18 |
| wallyworld | instead of html forms | 07:18 |
| wallyworld | so i extended the new subscribers_list portlet | 07:18 |
| wgrant | Right. | 07:18 |
| wallyworld | so, you ok with the change? | 07:19 |
| jtv | StevenK: still running | 07:19 |
| wgrant | wallyworld: Would be nice if we eventually integrated the same self-subscription stuff that Bugs introduced, but small steps. | 07:19 |
| wallyworld | wgrant: yes, one thing at a time. it's now much better than it was last week. it's now possible to unsubscribe without SQL :-) | 07:20 |
| wgrant | Yup. | 07:20 |
| * wallyworld off to soccer | 07:22 | |
| adeuring | good morning | 07:29 |
| allenap | Good morning/evening wgrant :) Got time for a very short review? https://code.launchpad.net/~allenap/launchpad/dsd-base-view-dont-modify-globals-bug-809985/+merge/70076 | 07:59 |
| wgrant | allenap: that's rather evil, but OK! | 08:01 |
| wgrant | Approved. | 08:01 |
| === wgrant changed the topic of #launchpad-dev to: Performance Tuesday | https://dev.launchpad.net/ | On call reviewer: - | Critical bugs: 241 - 0:[#######=]:256 | ||
| allenap | wgrant: Thanks :) | 08:02 |
| jelmer | g'morning | 08:09 |
| mrevell | Morning | 08:19 |
| === almaisan-away is now known as al-maisan | ||
| === al-maisan is now known as almaisan-away | ||
| === almaisan-away is now known as al-maisan | ||
| allenap | Morning jelmer :) | 08:43 |
| allenap | jelmer: I saw that Dulwich made the front page of Hacker News. | 08:44 |
| jelmer | allenap: yeah :) | 08:50 |
| jelmer | allenap: I've met the Google code folks and have been in contact with them. They've contributed a bunch of patches which should help with performance. | 08:55 |
| allenap | jelmer: Also, bug 808035 is at the top of the deploy queue and needs testing :) | 08:55 |
| _mup_ | Bug #808035: AttributeError: 'NoneType' object has no attribute 'write' during git import <oem-services> <qa-needstesting> <regression> <Launchpad itself:Fix Committed by jelmer> < https://launchpad.net/bugs/808035 > | 08:55 |
| jelmer | That should also have a significant impact on git code import performance. | 08:55 |
| jelmer | allenap: Yep, it's on the top of my todo list too. | 08:55 |
| allenap | jelmer: That's really cool :) | 08:55 |
| lifeless | adeuring: hiya | 08:56 |
| danilos | mrevell, hi, do you have any time today to help move the docs forward and potentially announce the change to maintainers by email (import queue stuff)? | 09:04 |
| mrevell | danilos, Yes, certainly. In 30 minutes. | 09:05 |
| danilos | mrevell, excellent, let's talk then | 09:05 |
| adeuring | hi lifeless | 09:10 |
| lifeless | adeuring: https://code.launchpad.net/~adeuring/launchpad/bug-739052/+merge/69625 - you're not landing it yet ? [I can't see the column issue addressed] | 09:12 |
| adeuring | lifeless: now running through ec2 | 09:12 |
| adeuring | forgot that yesterday... | 09:13 |
| lifeless | adeuring: and there were a bunch of suggestions that seem forgotten too :( | 09:13 |
| adeuring | I'm also working on the remaining suggestions you made | 09:13 |
| lifeless | adeuring: oh cool! - are you thinking of two separate landings ? | 09:14 |
| adeuring | lifeless: actually, three, to keep the diffs to review small | 09:14 |
| lifeless | sure thing; I'd be delighted to review those for you | 09:14 |
| lifeless | adeuring: have I mentioned how glad I am that you're doing this. | 09:14 |
| adeuring | lifeless: too late for this one : https://code.launchpad.net/~adeuring/launchpad/bug-739052-2/+merge/70044 ;) | 09:15 |
| adeuring | lifeless: and thanks :) | 09:15 |
| adeuring | (sorry, I have some reaally long lag between typing IRC messages and looking at incoimng...) | 09:16 |
| lifeless | heh no worries | 09:16 |
| lifeless | adeuring: so why won't (col1 >= M1 and col2 >= M2 and colN >MN) work ? | 09:18 |
| lifeless | oh, I see | 09:19 |
| lifeless | adeuring: however, using a tuple will work, I think ? | 09:21 |
| lifeless | hmmm, only if the direction on all the columns is consistent | 09:21 |
| adeuring | lifeless: no, I think we need, for three sort columns: | 09:22 |
| adeuring | col1 == memo1, col2 == memo2, col3 > memo3, then... | 09:22 |
| adeuring | col1 == memo1, col2 > memo2 | 09:22 |
| adeuring | then col1 > memo1 | 09:22 |
| lifeless | adeuring: what about | 09:22 |
| lifeless | (col1, col2, col3) >= (memo1, memo2, memo3) | 09:23 |
| adeuring | that wouldn't change much | 09:23 |
| adeuring | problem is: | 09:23 |
| adeuring | when the value in col1 increases, the values in col2, col3... starting from the minimum | 09:23 |
| adeuring | so we end up with N conditions for N sort columns | 09:24 |
| adeuring | lifeless: I think we have at least three options to "join" these N conditions: a more or less huge OR (implemented for mow), or a UNION ALL on sub-selects for each conditions | 09:26 |
| adeuring | or | 09:26 |
| adeuring | we can iterate in Python over the conditions | 09:26 |
| lifeless | adeuring: if the direction is consistent, a tuple works | 09:26 |
| adeuring | lifeless: ah, right! | 09:27 |
| lifeless | http://pastebin.com/mcShr1D2 | 09:27 |
| adeuring | lifeless: but how do we express this in SQL? | 09:27 |
| lifeless | just so | 09:27 |
| adeuring | lifeless: cool! | 09:27 |
| adeuring | so, the content of my next branch is obvious ;) | 09:28 |
| lifeless | :) | 09:28 |
| lifeless | we may have a lot where we can't do this | 09:28 |
| lifeless | but we can probably group all the adjacent same-direction columns anyhow | 09:28 |
| adeuring | lifeless: right | 09:29 |
| henninge | jml: Hi! ;) | 09:33 |
| jml | henninge: hi | 09:34 |
| henninge | jml: I got a failed ec2 run for your branch. | 09:34 |
| jml | henninge: me too, but no actual failures | 09:34 |
| henninge | jml: right | 09:34 |
| henninge | it's the "smtp message exeeds size limit" | 09:34 |
| henninge | jml: same for you? | 09:35 |
| jml | henninge: I've corrected the default value as per StevenK's comments. I suspect that would have caused some tests to fail. | 09:35 |
| jml | henninge: yes. | 09:35 |
| lifeless | that smtp error means a catastrophic failure | 09:35 |
| henninge | lifeless: you mean "too many failures, the mail gets too big" ? | 09:35 |
| lifeless | yes | 09:35 |
| henninge | jml: I see you pushed, shall I start the landing again? | 09:38 |
| jml | henninge: yes. thank you. | 09:38 |
| henninge | ok | 09:38 |
| lifeless | allenap: there is a feature flag you can use for profiling too | 11:03 |
| lifeless | allenap: the default profile dir is $cwd, I think. | 11:03 |
| allenap | lifeless: Oh neat. I was just shuffling my wiki notes; I think that one is from back in the days when BjornT_ ran the bugs team :) Not sure if it'll even work any more. | 11:05 |
| jtv | Any reviewers in the house? → https://code.launchpad.net/~jtv/launchpad/bug-819674/+merge/70135 | 11:09 |
| jml | who actually uses checkUpload via the API? | 11:11 |
| LPCIBot | Project devel build #938: STILL FAILING in 5 hr 44 min: https://lpci.wedontsleep.org/job/devel/938/ | 11:12 |
| jelmer | jml: I would imagine things like request-sync | 11:12 |
| jelmer | jml: to do things like determine if the current user requires sponsorship | 11:12 |
| jml | jelmer: ah, yeah, that would be a thing. | 11:13 |
| jml | for checking PPA uploads, it's not particularly useful | 11:13 |
| jml | since it errors when it doesn't recognize the source package, and insists that you supply component & pocket | 11:14 |
| jelmer | yeah, it's a bit odd | 11:15 |
| === jtv is now known as jtv-afk | ||
| lifeless | night all | 11:38 |
| jelmer | g'night lifeless | 11:38 |
| jelmer | jtv-afk: trade you for https://code.launchpad.net/~jelmer/launchpad/logginguifactory-fixes/+merge/70144 ? | 11:38 |
| StevenK | The publisher takes 4+ hours on DF? Srsly? | 12:14 |
| === matsubara-afk is now known as matsubara | ||
| bigjools | StevenK: it sometimes does | 13:08 |
| === mrevell__ is now known as mrevell | ||
| abentley | adeuring: Is it considered a feature that https://bugs.launchpad.net/mahara/+bug/630891%20/+index works? | 13:59 |
| _mup_ | Bug #630891: Unable to delete Notification messages <Mahara:Fix Released by richard-mansfield> < https://launchpad.net/bugs/630891 > | 13:59 |
| abentley | adeuring: i.e. a URL with whitespace in it? | 13:59 |
| adeuring | abentley: I have no idea... | 13:59 |
| abentley | adeuring: It seems bizarre, and it means that https://bugs.launchpad.net/mahara/+bug/630891%C2%A0/+index fails due to encoding issues. | 14:00 |
| _mup_ | Bug #630891: Unable to delete Notification messages <Mahara:Fix Released by richard-mansfield> < https://launchpad.net/bugs/630891 > | 14:00 |
| adeuring | abentley: I don't see how these two URL-oddities are related | 14:01 |
| abentley | adeuring: If URLs with spurious characters were not permitted, the second would give a LocationError instead of trying, and failing to render. | 14:02 |
| abentley | adeuring: i.e. a "Not found"/ "Lost something"? | 14:02 |
| adeuring | abentley: sure, a "not found" would be more reasonable, but the failure for your second URL is a separate issue: A unidoce decoding problem | 14:03 |
| adeuring | abentley: we would get that too for a project name like "mülltüte" | 14:04 |
| adeuring | but such names a not allowed | 14:04 |
| abentley | adeuring: Our URLs don't have non-ascii characters, so they can't have a unicode decoding problem. | 14:04 |
| abentley | adeuring: Except bug URLs, which can have spurious non-ascii characters. | 14:04 |
| adeuring | abentley: do we have a referer for the \xa0 / %C2%A0 URL? | 14:05 |
| adeuring | abentley: I mean: you can always type URLs manually | 14:05 |
| abentley | adeuring: No, but it was googlebot. | 14:06 |
| adeuring | abentley: so the URL has been somewhere/somehow generated. So we have two issues: the unicode error, and the fact that a spurious space at least in bug URLs does not result in a 404 | 14:08 |
| abentley | adeuring: Yes, that's my view. It hardly seems worth fixing the encoding bug if non-ascii URLs are supposed to be verboten. | 14:09 |
| abentley | adeuring: Who knows what other bugs non-ascii URLs might reveal? | 14:09 |
| adeuring | abentley: yeah,,, | 14:10 |
| adeuring | abentley: but regarding the working URL with the %20: | 14:10 |
| adeuring | Python 2.7.1+ (r271:86832, Apr 11 2011, 18:13:53) | 14:10 |
| adeuring | [GCC 4.5.2] on linux2 | 14:10 |
| adeuring | Type "help", "copyright", "credits" or "license" for more information. | 14:10 |
| adeuring | >>> int('1') | 14:10 |
| adeuring | 1 | 14:10 |
| adeuring | >>> int('1 ') | 14:10 |
| adeuring | 1 | 14:10 |
| adeuring | abentley: I guess that the travesal code does just such a simple int('630891 ') | 14:12 |
| abentley | adeuring: Interestingly: ValueError: invalid literal for int() with base 10: '2\xa0' | 14:12 |
| adeuring | abentley: try int(u'1\xa0') | 14:13 |
| abentley | adeuring: Fascinating. So that is only permitted for unicode strings, because only in unicode strings does \xa0 refer to the same codepoint consistently. | 14:16 |
| adeuring | abentley: doesn't the "can't encode character u'\xa0'" in the traceback indicate that dict['PATH_INFO'] already _is_ a unicode object? | 14:19 |
| abentley | adeuring: yes, that's what I think. | 14:21 |
| adeuring | abentley: so again not a "clear border" where 8-bit sequences with some encoding are supposed to live on one side and unicode objects on the other... | 14:22 |
| abentley | adeuring: Right. Requests appear to use a "sane_environment" where values are unicode, but converting the browser request to a web service request appears to re-convert the environment. | 14:24 |
| adeuring | abentley: so.... a trvial but bad workaround would be to convert to utf-8 in this case. | 14:25 |
| abentley | adeuring: Might be a necessary workaround, depending on how these calls are structured. | 14:26 |
| adeuring | abentley: if you want more fun, try this URL: https://bugs.launchpad.net/mahara/+bug/630891%C3%A4/+index | 14:27 |
| _mup_ | Bug #630891: Unable to delete Notification messages <Mahara:Fix Released by richard-mansfield> < https://launchpad.net/bugs/630891 > | 14:27 |
| adeuring | quite different error... | 14:28 |
| adeuring | %C3%A4 is a 'ä' | 14:28 |
| abentley | adeuring: Yeah, that's kind of error I'd expect. | 14:28 |
| adeuring | abentley: I assume that this error is raised for a failed int('630891ä') | 14:30 |
| adeuring | (for the bug id) | 14:30 |
| abentley | adeuring: me too. | 14:30 |
| abentley | adeuring: a0 is non-breaking whitespace. | 14:31 |
| adeuring | right | 14:31 |
| adeuring | so it might be a replacement from the regular '\x20' space | 14:32 |
| abentley | adeuring: Maybe. Since googlebot is the source, I imagine it's in some web page, though. | 14:33 |
| adeuring | abentley: right. And the s/\x20/\xa0/ might be deliberate: To avoid wrapping the URL... | 14:34 |
| adeuring | though as far as I know most webbrosers ignore the "don't wrap" for \xa0 | 14:35 |
| abentley | adeuring: but of course URLs shouldn't have whitespace. | 14:35 |
| adeuring | abentley: sure, but we can't stop the world mangling our URLs ;) | 14:35 |
| abentley | adeuring: we can, it's just hard to get venture capital :-) | 14:36 |
| adeuring | abentley: it might help to add a check like "strip(path_element) == path_element" before the int(path_element) | 14:36 |
| adeuring | ...where path element is the bug ID | 14:37 |
| abentley | adeuring: Yeah, that would work, or re.match('[0-9]*', path_element) | 14:37 |
| adeuring | abentley: right | 14:37 |
| adeuring | a nitpick: re.match('^[0-9]+$') | 14:38 |
| adeuring | abentley: for even more fun: https://bugs.launchpad.net/mahara/+bug/630891%E4/+index | 14:39 |
| _mup_ | Bug #630891: Unable to delete Notification messages <Mahara:Fix Released by richard-mansfield> < https://launchpad.net/bugs/630891 > | 14:39 |
| abentley | adeuring: what the?? | 14:39 |
| adeuring | abentley: %E4 is not vaild UTF8 | 14:40 |
| adeuring | cool, right? | 14:40 |
| abentley | adeuring: That's wacky. | 14:40 |
| === al-maisan is now known as almaisan-away | ||
| abentley | adeuring: merge proposals are also affected. | 14:58 |
| adeuring | abentley: interesting. Again via an integer ID? | 14:58 |
| abentley | adeuring: Right. | 14:58 |
| abentley | adeuring: I'm gonna guess that everything that has an integer ID is affected. | 15:00 |
| adeuring | abentley: seems this is worth a general helper function "check an int ID in a URL wfor all sorts of oddities" | 15:00 |
| abentley | adeuring: Yes, though I expect this kind of thing will rot. Maybe we should just reject URLs with whitespace from the beginning. | 15:03 |
| abentley | adeuring: Or more precisely, URLs with whitespace in the PATH. | 15:04 |
| adeuring | abentley: the you can also add checks for the other issues: any byte value outside the ASCII set | 15:05 |
| adeuring | more precisely: all bytes in the URL should have integer values between 33 and 126 | 15:05 |
| adeuring | ...127 | 15:05 |
| adeuring | ...after replacing the %xx encoding | 15:06 |
| adeuring | ouch.. the latter check should _not_ be done for the part after '?' | 15:07 |
| abentley | adeuring: Right, only the PATH. | 15:12 |
| adeuring | yep | 15:12 |
| === matsubara is now known as matsubara-lunch | ||
| LPCIBot | Project db-devel build #775: STILL FAILING in 5 hr 59 min: https://lpci.wedontsleep.org/job/db-devel/775/ | 16:56 |
| LPCIBot | Project devel build #939: STILL FAILING in 5 hr 45 min: https://lpci.wedontsleep.org/job/devel/939/ | 16:57 |
| === beuno is now known as beuno-lunch | ||
| jcsackett | sinzui: are you free to mumble? | 17:52 |
| sinzui | jcsackett, I am otp. I will be available in 45 minutes | 17:53 |
| jcsackett | sinzui: dig. | 17:59 |
| === beuno-lunch is now known as beuno | ||
| sinzui | jcsackett, I can talk now | 18:42 |
| jcsackett | sinzui: cool. | 18:42 |
| === matsubara-lunch is now known as matsubara | ||
| benji | Anyone want a small (and possibly even slightly fun) JS review? https://code.launchpad.net/~benji/launchpad/bug-809511/+merge/70219 | 20:48 |
| benji | sorry, gary claimed it so you all missed out | 20:53 |
| allenap | Hi, can anyone take care of reverting rev 13574 (from bug 810290). It's qa-bad, but I really have to go. | 21:39 |
| _mup_ | Bug #810290: no way to set scopes for incoming mail <feature-flags> <mail> <qa-bad> <Launchpad itself:Fix Committed by mbp> < https://launchpad.net/bugs/810290 > | 21:39 |
| lifeless | 13574 ? | 21:47 |
| lifeless | allenap: does it work *without* the feature rule in place ? | 21:49 |
| === matsubara is now known as matsubara-afk | ||
| LPCIBot | Project db-devel build #776: STILL FAILING in 5 hr 39 min: https://lpci.wedontsleep.org/job/db-devel/776/ | 22:35 |
| LPCIBot | Yippie, build fixed! | 22:46 |
| LPCIBot | Project devel build #940: FIXED in 5 hr 48 min: https://lpci.wedontsleep.org/job/devel/940/ | 22:46 |
| sinzui | StevenK, mumble? | 23:02 |
| wallyworld | sinzui: my sound is @!%@!%$!@^. rebooting | 23:09 |
| lifeless | matsubara-afk: wow, nice defect finding | 23:13 |
| === StevenK changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: StevenK | Critical bugs: 241 - 0:[#######=]:256 | ||
| poolie | hi all | 23:42 |
| wgrant | Hi poolie. | 23:54 |
| === wgrant changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: StevenK | Critical bugs: 248 - 0:[#######=]:256 | ||
| wgrant | poolie: Seen the latest on bug #810290? | 23:54 |
| _mup_ | Bug #810290: no way to set scopes for incoming mail <feature-flags> <mail> <qa-bad> <Launchpad itself:Triaged by mbp> < https://launchpad.net/bugs/810290 > | 23:54 |
| poolie | not yet, was just wondering about it | 23:54 |
| poolie | ah | 23:55 |
| poolie | sorry | 23:55 |
| poolie | so is that ok just rolled back, or is anything else needed? | 23:55 |
| wgrant | Should be all OK. | 23:56 |
| wgrant | Just takes 8 hours. | 23:56 |
| poolie | to roll it back out? | 23:56 |
| poolie | :/ | 23:56 |
| poolie | i did have some worries about whether this was sufficiently tested | 23:56 |
| wgrant | Yeah, because it missed buildbot by 3 minutes. | 23:56 |
| wgrant | So needs to wait for two runs. | 23:56 |
| poolie | i guess if it's trapping there the answer is that this code is not reached at all in the test suite | 23:57 |
| sinzui | wallyworld, is this what you were expecting? http://pastebin.ubuntu.com/657556/ | 23:57 |
| * wallyworld looks | 23:57 | |
| wallyworld | sinzui: yes, i think that's a bit nicer? | 23:58 |
| sinzui | The tests passed | 23:58 |
| sinzui | I will commit | 23:58 |
| wallyworld | \o/ | 23:58 |
| wallyworld | thanks | 23:58 |
| sinzui | wallyworld, Changes are pushed | 23:59 |
| wallyworld | thanks, will approve and organise a mentor | 23:59 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!