=== StevenK changed the topic of #launchpad-dev to: qastaging is down, but being worked on. |https://dev.launchpad.net/ | On call reviewer: StevenK | Critical bugtasks: 292 [00:38] Erm [00:38] r14367 is a little odd [00:40] Oh? [00:40] Naming cookies after the LP username. [00:40] Very odd. [00:41] qas looks okay to me, is the topic stale? [00:42] It's been back for an hour or so. === StevenK changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: StevenK | Critical bugtasks: 292 [00:42] Good enough for me. [00:45] lifeless: Was there consensus on which cookie approach to take? [00:50] wgrant: see the discussion on the MP that just landed [00:51] Hmm, OK. [00:51] Still seems pretty hideous to include the (mutable) username in the cookie. [00:52] agreed [00:53] deryck is coming around [01:28] wallyworld__, i guess feedback about the specifics of the disclosure ui would be beside the point? [01:29] Probably, but still useful. [01:29] poolie: for the email to the dev list yes. but huwshimi and mrevell etc would love to hear feedback [01:30] poolie: since we are iterating based on screenshots from 18 months ago and user testing now etc [01:33] poolie: also, that ui you saw from the email is now old. we're working on a new version, something like http://people.canonical.com/~huwshimi/policy7.png [01:34] ok [01:34] that looks a lot better [01:35] * poolie cancels his draft [01:35] maybe you can follow up with that? [01:43] fuuuuuuuuuuuuu [01:43] buildbot failed again. [01:43] * wgrant disables yuixhr [01:43] gary_poster: Around? [01:44] wgrant: I'm working on it [01:44] actual = ["Failure in lp/app/javascript/tests/test_lp_client_integration.Cache data.test_logged_in_user_has_cache_data: Unexpected error: Can't find variable: url", [01:44] wgrant: I thought you did that last night? [01:44] StevenK: I decided to give it another run to sort itself out. [01:44] It didn't. [01:44] wgrant: Or was that just a threat? [01:44] gary_poster: Any reason not to disable it now? [01:44] It's blocked all builds for more than a day now. [01:45] Ah, I see your email. [01:45] Thanks. [01:48] poolie: the ui from the screenshot is not quite ready yet. the point of the email wasn't get get feedback on the disclosure ui, but the technology used to produce the mockups [01:49] yeah i understand [01:49] it was just an example [01:49] i think it's great that you did an interactive mockup [01:51] it took a bit of work (lots of javascript hacking) but it's got some great feedback into the development of the feature [01:51] if we can speed up the process by which we can deliver such mockups and end up with less wastage (throwaway code), i think that will be great [02:07] i wonder if there is any path to having it running in an actual lp instance but with no backend [02:07] i don't know if that would actually be harder [02:11] poolie: pyramid or grok (ie a lightweight, easily deployable app server with zope support) i think may be easier [02:19] wallyworld__: is there any reason you can't just edit templates in-place on lp ? [02:19] wallyworld__: with the same persistence throwaway story ? [02:19] lifeless: wouldn't that involve landing stuff via pqm or whatever? [02:20] or do you mean ssh into the deployment server [02:20] and edit directly via the file system? [02:20] wallyworld__: is this up some place to play with? (seeing screenshot) [02:21] wallyworld__: ec2 demo [02:21] I'm curious what the edit link does, it seems like it'd just be on the person to overlay a ui to add/edit permsisions for that user vs per permission [02:21] rick_h_: the latest screenshot isn't (yet). only the earlier version as per the email [02:22] wallyworld__: e.g. just edit in your own local copy to hearts content, then shove it up on the ec2 instance (or canonistack or ...) [02:22] lifeless: ah, that may work. i haven't used ec2 demo before [02:23] rick_h_: the edit icon per person uses a ChoiceSource popup to edit the permissions for that person [02:23] As discussed on the call yesterday, ec2 demo probably doesn't quite work any more, but would be easy enough to fix. [02:24] wallyworld__: got it, missed the link when I read your email originally [02:25] the thing to remember also is that it has to be a RAD environment. hacking javascript and static html fits the bill there. but we want a bit more, like templates etc [02:25] wallyworld__: are you guys using a JS tpl language at all? [02:25] YUI is building in support for handlebar in their MVC stuff, but I've used jquery-tmpl and underscores stuff [02:25] rick_h_: no. but we would look to use mustache i guess since that's now in our technology stack [02:26] yea, I think deryck and the guys are using it on their feature landing [02:26] (it wasn't when we started the mockup) [02:26] sorry, handlebar/mustache [02:26] I group the two together when I think of them [02:26] sure :-) [02:27] I've always thought the couchapp stuff would be awesome for mockups, but I think it's a bit far from where you want to be [02:30] wallyworld__: small ui thing, curious why the "Add an Observer" wouldn't fit over on the right like a "Report a bug" or "Register a branch" type item? [02:31] * rick_h_ is checking out when different ui mechanisms are used like that [02:32] rick_h_: it may well go there. we were working off the supplied screenshots that were done a while ago [02:32] ah ok [02:32] poolie: rick_h_: latest version http://people.canonical.com/~ianb/disclosure/ [02:32] sorry, it's interesting stuff so I like to pester :) [02:33] I'd definitely try to go lighter on the grey for more contrast [02:33] but cool [02:33] rick_h_: please continue to ask questions. fwiw, a lot of what is there in the screenshots we are working of we agree needs to be fixed [02:33] so were are using them as a starting point [02:34] what would be sweet, would be if the grey boxes were buttons, and the edit link only showed on hover [02:34] that would clean it up a bit readable wise imo [02:34] That was what I was sorting of thinking, yeah. [02:34] The edit links all over LP are really cluttersome. [02:34] I know that hover stuff gets into the way of mobile, but I'm not sure how mobile friendly we are atm anyway [02:34] wgrant: yea, I want to align them something fierce lol [02:35] I've wished I had started back when the new buglist stuff had first started lol [02:35] Heh [02:35] Yeah [02:35] I like my lines through my ui/design I guess, though I hate css grids...so go figure [02:36] rick_h_: lighter grey now done :-) [02:36] ah, much nicer [02:37] so why can I click on some and not others? [02:37] * wgrant tries to dig up the faded edit icon. [02:37] rick_h_: atm, if you are in a team role, you get full observer rights (thats that the original screenshots called for). i'd have to check to see if that's still the case [02:38] s/team role/project role [02:38] wallyworld__: It's no longer the case. [02:38] ah, default perms then [02:38] It's merely the default during project setup. [02:38] gotcha [02:38] wgrant: np. i'll add the edit icons to those ones then [02:43] Aha, https://launchpad.net/@@/edit-grey [02:43] That's what I was looking for. [02:43] Slightly less abhorrent. [02:43] The initial inline bug status/importance used that until hover. [02:45] wgrant: i think anything is less abhorent than the yellow :-) [02:46] Surely not. [02:48] (I think ideally it would appear like a dropdown listbox, so it would have a down arrow instead of an edit icon. but lazr-js doesn't really do that sort of thing) [02:49] wallyworld__: I don't think it's useful to show "View nothing" rather than just hiding that policy. But then how do we add a new one. [02:50] wgrant: that's the dilemma [02:51] wgrant: we could have a proper popup form/widget [02:51] and a single edit icon in the table row [02:51] wallyworld__, are the pencils on the permissions really helping anything? [02:51] Our JS UI toolkit is sort of limited :/ [02:51] Yes, they're helping to lots of LP pages look crap :) [02:52] also, having the bug/branch column have no text is a bit odd [02:52] and the whole layout looks a bit oddly cramped [02:52] poolie: they are currently needed to allow the permissions to be edited [02:52] technically needed? [02:52] hm [02:53] poolie: the bug/branch column doesn't need text - it's the mechanism to navigate to the user's detailed disclosure page [02:53] wallyworld__: Well, might be useful to say "1234 bugs, 2 branches" or something like that. [02:53] exactly [02:53] and make that a link [02:53] it just looks broken at the moment [02:53] can do that [02:54] mm [02:54] to make the top of the thing less cluttered [02:54] Also, those are really an attribute of the "restricted observer" grant for a particular policy. [02:54] maybe move "add an observer" to the distinguished actions portlet in the top right? [02:54] like for 'report a bug' [02:54] "The list of observers can be large. Use Search to look for particular users. [02:54] The search term will be matched against name, Launchpad id, email. [02:54] " [02:55] seems like it doesn't really need to be there [02:55] can do that too. the current layout is as per the supplied screenshots that were were given to start from [02:55] it's pretty obvious what search will be for [02:55] wallyworld__: Are these mockups in a branch somewhere? [02:55] the heavy black rule through the middle of the filters is a bit odd too [02:56] might be better if all the filters were grouped within a box [02:56] "Show users which can view the project because they belong to a team which has permission." [02:56] wgrant: lp:~launchpad-purple-squad/+junk/disclosure_mockups [02:56] to start with surely you mean "who" not "which" [02:56] but [02:56] i don't know if that sentence is adding much [02:56] or what it's supposed to mean [02:57] poolie: the black line is to separate the controls since the ones above the line cause a xerver call, the ones below filter in place [02:57] :/ [02:57] ok [02:57] but rememeber, this is all exploratory, a starting point [02:58] sure [02:58] i'm exploring it :) [02:58] i think, if the line was not there [02:58] it would probably be clear that some of them act immediately and others need to be submitted [02:58] huwshimi: you may want to cut n paste the above discussion for feedback into the devel process [02:59] poolie: the line was added because users thought it wasn't clear [02:59] wallyworld__: I'll have a look [03:00] what specifically wasn't clear? [03:00] huwshimi: thanks. there's been some good feedback to add to whatever the user testers come up with [03:00] mm [03:00] wallyworld__: OK great [03:00] i think it's a bit like the green link thing [03:00] yes it means something [03:00] but is it something users actually need to know about? [03:01] poolie: it wasn't clear that the bottom controls acted in place vs doing a submit [03:01] if they need to know about it to use the thing effectively, it needs to be a distinction they'll actually understand [03:01] hm ok [03:01] poolie: i sort of agree with you. i was just following orders :-) [03:01] so they clicked it and [03:01] ... [03:01] trying to understand why that would be confusing [03:02] by that i mean i think it was indicated we need to distinguish between the two types of controls, but i could have mis remembered that [03:03] i don't think that's a goal in itself [03:03] the point is just that people can understand how to use it [03:03] so, so far we've gone from the original old screenshots to huwshimi's latest design which came from one round of user testing plus the ever evolving feature requirements [03:03] nobody else specifically distinguishes them so i'm really skeptical that we need to [03:03] the end users haven't yet seen this 2nd iteration [03:04] i think if we ask them "will foo cause a round trip to the server" they may not know but that's not a practical question [03:05] poolie: it's not so much a round trip to server thing vs a responsiveness thing. in-place obviously much quicker [03:05] but perhaps the distinction isn't required [03:05] mm [03:05] people will find out it's slow easily enough :) [03:06] like with the rest of lp untilthe recent performance improvements :-) [03:07] so, the idea now is to iterate to get to the point where the ui reflects the latest functional requirements. then we can bikeshed^H^H^H^H^H file tune :-) [03:16] wgrant: Can you have another look at https://code.launchpad.net/~stevenk/launchpad/use-userHasPriviledges/+merge/82967 ? [03:18] StevenK: Is DSP.driver necessary? [03:18] Also, I don't see any tests for SP/DSP.drivers [03:18] Apart from that it looks disturbingly sensible. [03:19] Disturbingly? :-/ [03:19] I'm not sure about DSP.driver, I don't think so, since the codepath checks if it implements IHasDrivers and then grabs .drivers [03:19] Right. [03:19] But you seem to have added DSP.driver. [03:20] And .drivers. I am a muppet. [03:21] wgrant: Right, binned. Let me write a test for both. [03:21] Thanks. [03:21] Should be pretty quick :) [03:24] wallyworld__: Which discussion exactly did you want me to look at? [03:24] huwshimi: just the feedback about the mockup which people saw after i sent the email to dev. eg search form, edit icons etc [03:32] wgrant: Diff updated [03:33] k [03:36] StevenK: Are you sure those new tests aren't just [] == []? [03:39] wgrant: Yes. I've added self.assertIsNot([], obj.drivers) where obj is the distro or the series and the tests still pass. [03:40] StevenK: k [03:40] Hah [03:40] wgrant: Do you want me to commit and push that change, or you don't care? [03:40] fhjwejiowefjio [03:40] StevenK: You know xfce4-smartbookmark-plugin? [03:41] How gina's been complaining about it for weeks? [03:41] StevenK: s/IsNot/NotEqual/, but yes. [03:41] Was that the non-free->main madness that the ftp-masters screwed up? [03:42] StevenK: No, it fails to unpack due to a patch referenced in series that doesn't exist. [03:42] It turns out that dpkg-dev is actually hideous. [03:42] And having reliably unpackable packages would be too easy. [03:42] There are vendor-specific series files. [03:42] Oh, right, that one. [03:42] So this unpacks fine on Debian, but not Ubuntu. [03:42] Bwaha [03:42] Because the bogus patch is mentioned in debian/ubuntu.series. [03:42] Sigh [03:42] Who could possibly think it was a good idea. [03:43] To make it ununpackable in that case. [03:43] So we need to patch dpkg to complain but not abort? [03:44] No, we probably need to slap whoever designed 3.0 (quilt) like that. [03:44] I'm not sure what dpkg-source can sensibly do here :/ === thumper is now known as thumper-afk [03:45] I wonder if --no-preparation works. [03:45] It does not. [03:45] Ah, --skip-patches [03:45] We do not care about patches for gina, surely [03:45] Exactly. [03:46] wgrant: Not so fast though -- does Lucid's dpkg support --skip-patches ? [03:46] I'm about to find out. [03:46] It does. [03:46] wgrant: Perhaps we should get wallyworld__ to fix it. Then he can touch Soyuz and wish he didn't. [03:47] I wonder what happens if I transfer the unpacked tree from an Ubuntu machine to a Debian one. [03:47] StevenK: Do you feel like filing a bug in Debian? [03:47] * wallyworld__ ducks for cover [03:47] Since I hate the BTS. [03:47] Against dpkg? [03:47] :-) [03:49] Hah [03:49] As much as I would like to consider the vendor-specific unpacking rules to be a grave bug, I don't think they'd like me much for that. [03:50] wgrant: Why don't you just configure reportbug? [03:50] STAB [03:50] poolie: wgrant: i added the grey edit icons [03:50] wallyworld__: I doubt they look good, but they're probably less offensive. [03:50] * wgrant checks. [03:51] Hah [03:51] The circle is just slightly visible. [03:51] less offensive is also what i'd say [03:51] wgrant: Those two extra asserts are in the diff. [03:51] But still looks much better. [03:51] yeah [03:51] With slight tweaking of the (otherwise unused, AFAIK) icon, it could look quite reasonable. [03:51] In fact, possible just drop the circle entirely. [03:52] StevenK: lol [03:52] StevenK: dpkg-source: warning: --skip-patches is not a valid option for Dpkg::Source::Package::V3::native [03:52] StevenK: So it warns if you ask for reliable extraction all the time. [03:52] Fucking hell. [03:53] It still works. [03:53] But spews warnings. [03:53] wgrant: However, here it unpacks, just exits 2 [03:53] StevenK: Ah, is that so... [03:53] I didn't think about that. [03:53] I wonder what 2 means. [03:54] I wonder if I can claim serious [03:54] important, at the very least [03:55] * wgrant watches for the wontfix [03:56] * wgrant wonders why we don't have any running water. [03:56] * wgrant goes hunting. [03:58] wgrant: debian/patches/series is a symlink to ubuntu.series [04:01] StevenK: In Ubuntu's version, or Debian's? [04:01] Debians [04:02] So Debian's xfce4-smartbookmark-plugin has a patch applied to direct people to launchpad to file bugs [04:02] Rather than the BTS [04:03] Which version? [04:04] In 0.4.4-1, debian/patches/series doesn't exist. [04:04] Because Debian's patches were included upstream. [04:04] I have 0.4.4-1 unpacked, and it's a symlink [04:05] Oh, that's dpkg-source being stupid [04:05] Hah [04:06] Nastygram filed [04:07] wgrant: Now that I've wrangled the BTS for you, can you wrangle my MP for me? [04:08] Sorry, got distracted by lack of water. [04:08] And dpkg-dev [04:16] wgrant: Thanks, tossing at ec2. [04:23] wallyworld__, grey just looks kind of disabled [04:23] yeah, i guess so [04:24] is there a technical thing that will break if there's no icon? [04:24] but final styling will come later [04:24] or is it more just that we think people won't realize they can click it...? [04:24] you get a hover underline [04:24] that makes it pretty obviously clickable to me [04:24] i think people expect an icon [04:24] huwshimi, remove bug tags woo [04:25] but i'll defer to the usability testing for what we need to do [04:25] yeah, me too [04:25] i just want the possibility of no pencils to be considered [04:25] * StevenK stabs jtv [04:25] the argument is basically just "we've always done that" [04:25] What have I done now? [04:26] Or are you just exploring different uses of pencils? [04:26] jtv: Inflicted that damnable buildd-manager branch on me [04:26] haha [04:26] SCORE! [04:26] I was going to mention that. [04:26] At some point. [04:26] Almost certainly. [04:27] Cross my heart and hope to be just outside my home when a slight drizzle starts. [04:27] wallyworld__, so while we're waiting for the testing [04:27] i don't know, is there any other reason than just people might expect it? [04:28] jtv: We've had 23mm here since 9am [04:28] .. i don't want to be a pain [04:28] i just really think it looks bad without adding anything [04:28] we have to stop some day [04:28] poolie: well, i would think we need to be consistent with the test of lp [04:28] s/test/rest [04:28] wgrant: Debian #649680, since the BTS sucks. [04:28] <_mup_> Bug #649680: Link titles of pages that aren't UTF8 appers as gibbrish < https://launchpad.net/bugs/649680 > [04:28] cause the rest of lp has totally consistent ui? [04:28] :) [04:28] mostly [04:29] poolie: I thought you knew! [04:29] more or less [04:29] perhaps [04:29] I think we need some indication that it's editable. [04:29] consistency has value but [04:29] A pencil is probably not it. [04:29] if you want to change things [04:29] It's really a list both. [04:29] List box. [04:29] you've got to start somewhere [04:29] sure [04:29] so one pattern people use is a little [v] icon when it pops up a control [04:30] especially a list [04:30] Exactly. [04:30] That's exactly what this is. [04:30] But lazr-js doesn't really do that sort of thing. [04:30] but really on many sites just hover underline/color/background seems to be enough [04:30] It prefers invoking a popup somewhere else on the page. [04:30] it invites people to click and then they can see what happens [04:31] my thinking is that lp is being rebranded atm, so we should wait for that to establish the guidelines [04:31] we have other things to deliver [04:31] fair enough [04:32] poolie: but i'm sure huw is open for suggestions on his rebranding work :-) [04:34] :) [04:52] StevenK: thank you for your sage comment on my branch. [04:53] The bit about the vodka was particularly insightful. [04:53] I think you're right. [04:53] About which bit? [04:54] The vodka. [04:54] the vodka, clearly [04:55] wgrant: do you drink vodka? I'm asking because we need another expert opinion on https://code.launchpad.net/~jtv/launchpad/bug-717969/+merge/83022 [04:56] A branch about vodka? [04:56] * nigelb looks [04:57] nigelb: nooooooo! Don't! [04:58] Poor man. Went in without his Absolut goggles. He'll never be the same. [04:58] Hahaha. Nice. [04:58] I stayed away from the code. So not scarred. [04:58] The only read your description and StevenK's rerview. [04:58] *review [04:59] * jtv tearfully greets nigelb back [05:00] :) [05:01] Yay [05:01] UnicodeDecodeError in bson.loads [05:02] LIFELESS [05:03] ooh. Caps Lock. [05:05] StevenK [05:06] lifeless: I was just guessing the Unicode issue was from oops-tools or something [06:52] hm so merging this loggerhead export patch raises the question of upgrading loggerhead as a whole [06:59] same as other source deps - land a change to LP to upgrade it === thumper-afk is now known as thumper [08:34] ok [08:34] tarball download works here; it should be ok to get it in tomorrow [09:11] Morning === almaisan-away is now known as al-maisan === matsubara-afk is now known as matsubara === al-maisan is now known as almaisan-away === matsubara_ is now known as matsubara === almaisan-away is now known as al-maisan [11:55] morning [12:18] Good morning rick_h_ [13:58] wallyworld__: wgrant is there a branch for the disclosure stuff I can see/find to checkout? === matsubara is now known as matsubara-lunch [14:34] sinzui: hola, I noticed your branch ~sinzui/html5-browser/trunk doesn't seem to exist anymore, which one should I be using instead? [14:34] ! [14:34] I will find it and put it back [14:35] sinzui: thanks man, it's been really nice for testing yui [14:38] cr3, I am helping another user. It will be a few minutes [14:39] sinzui: no worries, take your time [14:48] cr3, can you see https://code.launchpad.net/~sinzui/html5-browser/trunk ? [14:48] I just branches lp:html5-browser [14:50] I just branched lp:html5-browser [14:50] sinzui: seems to work, but it didn't work when building overnight: https://launchpadlibrarian.net/85743890/buildlog_ubuntu-precise-i386.launchpad-results_0.13-1%7Eppa0.12.04_FAILEDTOBUILD.txt.gz [14:53] ah. cr3, I think gary was changing lp's buildout rules. That was added recently. I created debs, but buildbot does not have them [14:54] sinzui: the package seems to have built fine for 10.04 and 10.10 though, only later releases failed. anything I should do? [14:54] rick_h_: I don't see assert_change being used anywhere. Is it dead code? [14:54] looking [14:55] yes, changed to using the .wait() thanks [14:55] cr3. I will copy the package to precise. I may want to do that for a few other too [14:55] sinzui: the build also failed for 11.04 and 11.10 though [14:56] I am not sure what to say, maybe gary_poster can help since he added the source deps [14:56] what did I do? [14:57] All I remember doing was adding html5_browser as a buildout dependency to lp [14:57] gary_poster: can you have a look at this build failure, seems to be related to buildout: https://launchpadlibrarian.net/85743890/buildlog_ubuntu-precise-i386.launchpad-results_0.13-1%7Eppa0.12.04_FAILEDTOBUILD.txt.gz === abentley changed the topic of #launchpad-dev to: https://dev.launchpad.net/ | On call reviewer: abentley | Critical bugtasks: 292 [14:57] but I didn't touch the debs [14:59] sinzui, cr3, this seems to be the heart of the issue: http://pastebin.ubuntu.com/747089/ [14:59] I don't think I touched html5 browser's buildout either, though I could be wrong [15:00] but I don't think that error has anything to do with buildout anyway [15:00] gary_poster, who put html5 in buildout? I only make deb packages [15:01] gary_poster, sinzui: I'll try rebuilding in the ppa and then maybe try locally with sbuild [15:01] sinzui: I put html5 in my buildout [15:01] ah, thank you for clarifying [15:01] sinzui, I put html5 in LP buildout. That has nothing to do with this error [15:03] gary_poster: so we both have html5 in our buildouts, do you have a tarball in your sourcedeps branch or do you branch html5 directly in buildout? [15:03] cr3 tarball [15:03] gary_poster: maybe I should do the same... [15:13] rick_h_: It's really helpful if you include my nick. This can be a noisy channel, but using my nick makes your replies stand out. [15:13] abentley: yea, sorry. [15:14] I forget that not all irc has nick highlighting and such. I know some do notifications so try to avoid doing that too much [15:14]