[00:19] poolie: Bug tag list is live :D [00:20] poolie: I know you've only been waiting 10 months for me to do it :) [00:22] lol, congrat huwshimi [00:22] where can I see huwshimi ? [00:23] rick_h_: heh, thanks :) [00:23] rick_h_: https://bugs.launchpad.net/launchpad/ [00:23] is it the tag list on the right? [00:23] with the counts? [00:23] rick_h_: Yeah [00:23] huwshimi, that's awesome [00:23] it doesn't look like a cat threw up :) [00:24] rick_h_: Not sure if you noticed, but they used to look like the left version of this: https://launchpadlibrarian.net/62993362/tag_cloud.png [00:24] ah, ouch [00:24] Hmm. [00:25] Interesting. The sizes were meant to scale more than that. [00:25] But it shows them all as identical... [00:25] wgrant, in the 'current'? [00:25] they did scale [00:25] That's what I thought. [00:25] huwshimi: now what about client side filtering box? :) [00:25] it did not work well because ... [00:25] well for a bunch of reasons [00:25] wgrant: They did scale, but the top tags were all a similar size [00:26] yea, those types of tag clouds never lookd great, especially with large phrases for tags [00:26] one of them being the words are so wide relative to the portal [00:26] mm [00:26] keeping the same size and varying intensity would be better [00:26] would have been [00:26] Yeah. [00:26] just showing the number is pretty good though [00:26] so huw while you're here [00:26] https://code.launchpad.net/ has another odd (and woefully inconsistent with the rest of the site) cloud. [00:26] rick_h_: Doing the tag filter would be nice, but this was an easier win as a first stage [00:27] wgrant: Yeah, from memory all those landing pages have something similar [00:27] huwshimi: oh yea, just helping you with stage 2 :) [00:27] wgrant, i wonder if we should just redirect that to the top home page [00:27] Amusingly enough it also doesn't include distributions, so Ubuntu doesn't show up. [00:27] huwshimi: No, that's the worst bit. Only code does. [00:27] maybe look at the logs to see if it is used [00:27] wgrant: Oh really? [00:27] i think the idea was that it would show activity [00:27] well, that was the idea [00:27] it was my idea [00:28] it is not actually doing the job very well [00:28] :/ [00:28] poolie: needs a color scale as well as size maybe? [00:28] there are two mismatched shades of blue on that page too [00:28] Some of them used to be green. [00:28] IIRC those that didn't use codehosting officially. [00:28] wgrant: it seems you're right [00:28] Can't quite recall, though. [00:28] https://code.launchpad.net/ also has the 1.0 buttons. [00:28] Ew [00:29] Tempting to remove the subdomains tomorrow, but that might be a bit controversial. [00:29] heh, #03a and blue [00:29] wgrant: yeah a few of the landing pages have those buttons [00:29] wgrant: on a saturday? yes, 'might' be [00:30] hey, it's friday [00:30] wgrant: Actually all but bugs do [00:30] always good to change right before the weekend and then hide [00:30] rick_h_: O hai. You haz QA. [00:31] StevenK: yep, am I holding up anything? Been afk for the holiday today. [00:31] wgrant: I'm definitely buy a beer for anyone who removes subdomains [00:31] rick_h_: There's only three revisions, in the queue. [00:31] s/,// [00:31] The only sticking point is the breadcrumbs. [00:31] rick_h_: So I don't care much. [00:31] StevenK: ok if I get it in the morning then? [00:31] Everything else is mostly a matter of deleting code. [00:32] StevenK: ok awesome, will hit it up first thing [00:32] rick_h_: If you're at work on your Friday, that sounds great. [00:32] StevenK: yep, will be. New guys don't have all that banked vaca :) [00:32] Haha [00:32] wgrant, getting everything on one domain so that it can run from a single ip without /etc/hosts futzing would be awesome [00:32] wgrant: you know, I've used LP on and off for years and only *just* noticed the breadcrumbs the other day [00:33] The breadcrumbs have been through a few iterations. [00:33] caught me totally out of the blue [00:33] This last variety has been in place for nearly 2.5 years, and is extremely subtle. [00:33] And confusing to an extent. [00:33] poolie: Indeed. [00:34] That requires ripping out vostok [00:34] Which might make mwhudson sad. [00:34] There's a branch for that.™ [00:34] <3 I think I have a new email sig [00:34] * huwshimi contemplates pie incentive to have subdomains removed by Budapest, or the end of Budapest [00:34] StevenK: vostok can be deleted [00:35] what's a vostok? [00:35] vostok is Linaro's aborted attempt at making a standlone Soyuz. [00:35] poolie: we can't run on one domain safely, ever. But we can shrink the number :) [00:35] * rick_h_ is grepping around too, I know I've run into that [00:35] lifeless: Well. [00:35] StevenK: I was discussing nuking vostok with mwhudson the other day. [00:35] lifeless, you're talking about librarians, ppas, etc? [00:35] lifeless: The webapp can be on a single domain. [00:35] wgrant: launchpadlibrarian [00:35] And that's the main thing. [00:35] ok [00:35] launchpadlibrarian is not user-facing. [00:35] In the important sense. [00:35] wgrant: internalxmlrp [00:35] perhaps for demo purposes it would be ok to have it on the same domain [00:36] wgrant: and the public xmlrpc too [00:36] wgrant: and apis [00:36] if there is no untrusted or important data [00:36] huwshimi, while i've got you [00:36] poolie: Yup, here [00:36] In dev I imagine they would just run on different ports. [00:36] some time i'm going to land the initial markdown thing [00:36] for people and products [00:36] i guess i feel i want a thumbs up/down from you [00:37] wgrant: perhaps; I'd be keen for dev to just use ip addresses for all the services. [00:37] lifeless: Sure, but that requires lots of config unless you run everything on loopback. [00:37] Which is limiting. [00:37] poolie: A massive thumbs up for the spirit of it, is there anything in particular you want validation on? [00:37] wgrant: why ? [00:37] lifeless: My machine has one non-loopback IPv4 address. [00:37] My VMs, too. [00:37] just that the list of 'do later' things on the bug and mp are ok [00:37] wgrant: ...and ? [00:38] lifeless: We can't run services on separate IP addresses if there's only one IP address. [00:38] StevenK: where do you go to see the pending QA and such? [00:38] huwshimi, the other thing is [00:38] rick_h_: The deployment report. [00:38] wgrant: I didn't intend to imply separate ip addresses [00:38] rick_h_: http://lpqateam.canonical.com/qa-reports/deployment-stable.html [00:38] rick_h_: Know it, love it. [00:38] i was contemplating, when my current lp discretionary hacks are finished, doing a really cheap timeline that just shows all the comments on bugs i'm subscribed to [00:38] lifeless: Ah, so separate ports, like I said. Sounds good, and is what I had planned. [00:39] in reverse chronological order [00:39] wgrant: and no domains [00:39] this seems [00:39] Sure. [00:39] wgrant: so an ip address on ec2 just works. [00:39] possibly useful [00:39] StevenK: thanks, bookmarked [00:39] I have 5 branches here relating to cleaning up the vhost machinery and config. [00:39] a toe in the water; and doesn't require db or infrastructure changes [00:39] poolie: timelines.... thanks for reminding me. I want notification to be done as a service. [00:39] rick_h_: There is also db-stable.html for database stuff [00:39] I should probably merge and land them at some point. [00:39] IIRC one of them destroys vostok. [00:40] lifeless, that sounds good but what exactly do you mean [00:40] poolie: I'm not 100% sure what I mean yet :). [00:41] blah [00:46] lifeless, ok :) [00:46] so like i say this seemed like a place to start [00:50] lifeless, StevenK: sure, rip out vostok if it's causing any problems [00:52] poolie: What happens when you dump all different types of content (headings etc) into a description, does it come out looking alright? [00:53] wallyworld_: what particular spinner did you have in mind in your review? [00:54] abentley: there isn't any additional spinner on the page right now, but there could be one added later and it could cause problems. so the suggestion is just to make the code more robust [00:55] but if you don't think it's required, i can just +1 it [00:56] it seemed worth making the change thogu, given it's just a slightly longer select expression [00:56] wallyworld_: There won't be a later. Deryck is working on a different spinner that will replace this one. [00:56] ah, ok. in that case i'll just +1 it [00:58] huwshimi, in markdown? [00:58] it looks ok [00:58] wallyworld_: thanks. [00:58] abentley: done, sorry for the confusion on my part [00:58] poolie: Yeah, ok great [00:58] obviously if people start putting lots of h1s it will look a bit weird but it doesn't spoil things [00:59] ... i don't think we're going to allow the animated gif background tag :) [00:59] poolie: Great, in that case I'm happy to go with what you have [00:59] poolie: Awww [00:59] rupert murdoch can buy lp for $580M [01:03] poolie: With the cloud on the code page I wonder if we can turn that into an activity wall of sortts [01:03] *sorts [01:04] That page is probably the walliest thing in LP right now. [01:04] It has the registered/changed branch feeds at the bottom. [01:04] And the cloud showing activity levels. [01:04] huwshimi, i don't know if you have time but if you want to do a sketch of a general activity wall , version 0 [01:04] i might try to implement it [01:06] poolie: I started working on this a long time ago: http://people.canonical.com/~huwshimi/profile_01.png [01:07] poolie: But I'm happy to work on this a bit with you [01:07] poolie: It would give us a chance to try a few thigns [01:07] *things [01:15] poolie: I'm just running your markdown branch, I've enabled the feature flag ("markdown.enabled default 1 true"), but I'm not getting the markdown rendered. What else might I need to do? [01:15] poolie: Oh, all I did was grab a new lp branch, and merged your changes [01:19] poolie: so I think what I mean is a few things that all intersect [01:20] poolie: notification / walls / timelines all belong outside the main lp codebase: it requires a rather different view of the data than our main tables are structured for [01:21] poolie: it also has no requirement to be carefully modelled - in fact it has a requirement to be agile and loosely modelled, to easily be updated as other things change [01:35] huwshimi, did you work it out with md? [01:35] poolie: erm, no I got distracted by other things :) [01:36] huwshimi, i don't know, that should be all you need to do [01:36] nb it only works on eg person descriptions, not on all text yet [01:36] lifeless, i agree with those as an ultimate goal [01:36] mm, not quite sure about why it ought to be outside the main codebase though [01:36] except in as much as perhaps it would be nice to have a less monolithic codebase generally [01:37] poolie: Oh, I misunderstood (I thought it worked on bug descriptions too). Yes it's working perfectly [01:40] just being (perhaps over) cautious [01:40] also deferring making it work with edit widgets [01:40] http://people.canonical.com/~huwshimi/profile_01.png looks good [01:40] i will try for something like that [01:43] poolie: multiple services will need to talk to notifications; its easier to work on small codebases; nosql may well be much easier to work with; all new things should be built with SOA in mind [01:44] k [01:45] notification/walls/timelines are greenfield, making them 5-10 times harder by working within the big curly DB we have would be a costly choice IMNSHO [01:45] ! [01:45] interesting [01:45] [email notifications clearly aren't greenfield, the rest are] [01:46] if they were done externally, how would the web content be visible to the user [01:46] in LP you would have a thin shim which makes API calls [01:46] same as e.g. talking to memcached, but without the potatoes. [01:49] wall_service.get_wall_html('mbp')? [01:49] for instance. [01:49] UI and rendering is in LP [01:49] ah, so more like [01:49] storage, graph traversal, summary snippets and a pointer back to the source document are in the service [01:50] the view object makes a call to get structured machinereadable data from the wall service? [01:50] yup [01:50] Privacy is a bit of a challenge, but not insurmountable. [01:50] all modern sensible stuff and isolated from insane http headers [01:50] so json blob [01:50] right, it will need to get it on behalf of the current interaction user [01:50] some implementation choices around whether the summary snippets are served as is or privacy-checked on the LP side etc [01:51] mm [01:51] i feel like i would like to do this in a way that for version 0 talks to the lp database [01:51] but with a view to eventually pointing to something else [01:51] you're welcome to scrach that locally. [01:51] but [01:51] quite isolated [01:51] I'm going to reject landing anything in this space in lp proper [01:51] :( really? [01:52] I realise this provides a barrier to entry, but I think its better overall [01:52] yes, really. [01:52] the cost of setting up a microservice is fairly modest [01:52] lol [01:52] ok [01:52] We are yet to have a single microservice running. [01:52] maybe i'll test that [01:52] (but I agree it needs to be done as one) [01:52] wgrant: not true; we have 3 running so far - txlongpoll, oops-tools and arguably rabbit [01:53] txlongpoll and rabbit are not running. [01:53] wgrant: txlongpoll only needs the rabbit config switched on in LP, and thats waiting for red to put the MP forward [01:53] They're somewhat close. [01:53] But they're not running. [01:53] so every lp action that does a wall-related activity will post it to the microservice? [01:53] rabbit is [01:53] and they'll be stored redundantly there? [01:53] lifeless: It's not being used for anything. [01:53] graphs and nagios are in place for prod rabbit [01:53] poolie: yes [01:53] poolie: for whatever depth of history we want to support [01:53] hm [01:54] obviously there are consistency things about [01:54] comments being deleted or hidden [01:54] for instance [01:54] this is covered under the question 'some implementation choices around whether the summary snippets are served as is or privacy-checked on the LP side etc' [01:54] but yes [01:54] there are [01:57] hm [01:57] i think a soa would be better [01:57] it seems a shame to block changes that give better presentation of data that is already in the db [01:57] perhaps it is necessary if things are ever going to improve [01:58] so, you mentioned, for instance, listing all your own comments on bugs, in date order; this requires joining all messages from $user, across to all bugs, then checking for privact [01:58] a few users have > 100K messages [01:59] I can pretty much guarantee you'll have timeout issues in any first-draft implementation on e.g. ~janitor [01:59] sure [01:59] but https://bugs.launchpad.net/~pitti/ eg times out today [01:59] this isn't a reason not to do it, but its a demonstration of how our schema is not arranged to meet the needs of that project [01:59] most bug list pages are timing out [02:00] garh [02:00] we fixed ~pitti [02:00] thats a regression [02:00] ~mbp times out [02:00] so [02:00] i don't want to add more timing out pages [02:00] great! :) [02:00] but i guess, [02:01] lp does have to have a general answer to "do simple select/sort queries across the bug db" [02:01] which is all this would need [02:02] i'm not really confident that bringing up a separate db that will scale to that size will be easy [02:02] poolie: Did you really prefer a 6 digit bug to a 4 digit one? [02:02] (even if it's allowed to throw stuff away) [02:02] yes, becaues it has a better description [02:02] but i don't care if you reverse it and update stuff [02:03] lifeless, anyhow it's kind of blue sky now [02:03] so i don't want to distract us with it [02:03] i support soaization [02:03] i don't want to add more timeouts [02:03] i do want to take a small step towards timelines [02:03] when i get to it i'll see if i can resolve those things [02:12] Launchpad is fixed! [02:12] * 14 Exceptions [02:12] * 2 Time Outs [02:12] Our work is done. [02:12] ? [02:13] lifeless broke oops-tools, so the OOPS reports aren't as depressing as usual today. [02:13] well there is certainly *something* odd ;( [02:44] hi, could i get a quick scan of https://code.launchpad.net/~mbp/loggerhead/240580-tarball/+merge/83364 from someone? [02:55] thanks lifeless [02:56] jtv: you dropped off [03:07] poolie: got some time for bouncing ideas around ? [03:10] lifeless, sure [03:10] can i have some lunch first? [03:10] sure [03:15] rabbit just failed again; i'm glad i fixed the feature :) [03:15] psa [03:29] Grumble, storm validators suck [03:35] wgrant: Can haz help? [03:35] whats up [03:36] IProduct.{_owner,security_contact} both have a storm_validator of validate_person_or_closed_team. [03:37] Setting the owner to an open team results in an error of "Constraint not satisfied", but setting the security contact to an open team gives "You must choose a valid person or team to be the security contact for " [03:38] And both validators are the same, so I'm not sure what is going on. [03:41] is there a try:except: around the other case ? [03:41] lifeless, hi, now? [03:42] ah i remember trying to figure out where constraint not satisfied came from [03:42] sure [03:42] pots? [03:43] lifeless: They're both in the same view, neither are checked in the view's validate() [03:44] StevenK: storm_validator is in the model code, i bet _owner has something funky in the zope.schema/interface side [03:45] mwhudson: There's a owner setter/getter, do you think that is interferring? [03:46] StevenK: no, i think the error is probably coming 'before' that [03:46] although owner is just an Attribute? [03:47] It's a PersonChoice [03:47] ah yeah [03:47] vocabulary='ValidPillarOwner', [03:47] mwhudson: The error isn't coming from the vocab, though [03:48] and i see security_contact is the same anyway [03:48] StevenK: only difference i see is PublicPersonChoice vs PersonChoice ? [03:49] Oh, which is done via IHasSecurityContact [03:49] StevenK: constraint not satisfied is raised in the form validation code in some frustrating way that blocks you from finding out why [03:50] yeah [03:50] if value not in vocabulary: [03:50] raise ConstraintNotSatisfied(value) [03:50] Ah ha [03:50] Is that in zope guts? [03:50] zope.schema-3.5.4-py2.6.egg/zope/schema/_field.py [03:50] RARGH [03:51] STEVE SMASH [03:52] mwhudson: I don't like the idea of rummaging around in self.errors in the view's validate() just for a better error. [03:52] StevenK: i can't remember what i did last time i got angry about this [03:53] StevenK: i'll notice that security_contact gives a nice error, maybe you can see why? :) [03:53] oh [03:53] well, slightly nicer [03:53] one that's from the launchpad tree at least :) [03:53] mwhudson: That's what I'm trying to figure out! :-) [03:54] StevenK: heh [03:54] StevenK: i hope this pointer helped a bit at least [03:54] The ConstraintNotSatisfied exception is in self.errors for owner when it's an open team [04:01] Right, there is a bunch of machinery around security contact [04:01] Whereas there is none around owner [04:11] mwhudson: I've added a block before the OMG-Zope-gave-us-errors-must-return section and that works. [04:16] StevenK: i'm not sure what that means, but i'm glad it works :) [04:17] mwhudson: http://pastebin.ubuntu.com/748928/ [04:18] StevenK: ah [04:20] * StevenK shakes fist at NameAlreadyTaken [04:33] wallyworld_: Hey mate, would you be around in about an hour to have a chat about getting the manage-disclosure mockups ready for testing? [04:35] huwshimi: i have to go to the post office quickly after school pickup so i'll be a little later than normal getting home. i'll be back around 4:45 AEDT [04:35] wallyworld_: Sure, that's fine [04:36] excellent. i'll ping you [04:36] wallyworld_: Cheers [05:37] huwshimi: ping :-) [05:46] wallyworld_: Hey, mumble? [05:46] huwshimi: yep, assuming it starts this time without locking up :-( [05:47] wallyworld_: skype, phone etc. are fine too if need be [05:49] why is lp.dev giving me a sad face? [05:50] oh ok now [06:05] i'm going to send through the update to loggerhead in the lp tree === wallyworld_ changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: - | Critical bugtasks: 292 [07:26] huwshimi, i wonder if we could usefully autosave in progress typing into local html5 storage [07:26] perhaps if your browser crashes it will be lost anyhow [07:26] i don't know [07:27] maybe it's the browser's job not to crash :) [07:28] wgrant: help, I'm stuck! [07:28] wgrant: remember bug 876594? [07:28] <_mup_> Bug #876594: rejected builds for synced packages send mail to Debian maintainer

< https://launchpad.net/bugs/876594 > [07:29] I'm trying to reproduce the problem in a test, but I need something like 4 layers of objects that we don't have LaunchpadObjectFactory methods for. [07:33] Or 5. [07:33] add em? [07:33] Yak shaving weekend? :) [07:34] personally i'm doing all the bzr reviews [08:01] is there a tasteful place to hold an object for the duration of the current request? [08:21] jtv: SoyuzTestPublisher [08:21] poolie: No, but there are some distasteful ones. What do you want to store? [08:21] re https://code.launchpad.net/~mbp/launchpad/391780-markdown/+merge/82832 i was going to cache the markdown converter object for later reuse [08:22] but, it turns out not to be all that expensive to just make a new one each time its wanted [08:22] so, nm [08:24] Ah [08:28] how does the 'incremental fix' thing work? [08:28] multiple fixes for the same bug. I think the flag is --incremental for ec2land [08:28] yep [08:28] The bug isn't set Fix Committed. [08:28] but you get the qa tags on it. [08:29] that sounds like just what i want then [08:29] i'm sending up the markdown branch as a limited beta [08:29] ahh. Nice. [08:29] Also, the new Beta thing is pretty cool. [08:30] wgrant: that's the answer I feared… I hope SoyuzTestPublisher gives me a say over the changes file. [09:12] Hello === almaisan-away is now known as al-maisan [09:41] nigelb: poolie: actually, thats not what happens [09:42] the qa process wiki page covers it I think; your revision will be skipped [09:42] is qas not updating or something? [09:42] i have something landed a while ago that's not on http://lpqateam.canonical.com/qa-reports/deployment-stable.html [09:54] poolie: A while ago? [09:54] poolie: Has it been through buildbot yet? [09:56] poolie: Thanks for looking at https://bugs.launchpad.net/launchpad/+bug/894045. I was on quite the wrong tack! [09:56] <_mup_> Bug #894045: sending pgp inline in plain text causes apparent crc errors < https://launchpad.net/bugs/894045 > [10:03] Where did the statuses on the translations-import-queue pages go? [10:07]