[00:04] <pgquiles> maxb: I was still around. debootstrap --variant=buildd, then running the root/bootstrap script almost gets it. I needed to modify the script to add 'universe' to etc/apt/sources.list and to create user and group buildd. I'm still missing 3-4 .conf files but I'll continue tomorrow. I've e-mailed lamont my findings and waiting for his answer.
[00:06] <wgrant> pgquiles: LP will override sources.list; you shouldn't need to add anything to it.
[00:06] <maxb> pgquiles: thanks. looking forward to what you hear
[00:08] <pgquiles> wgrant: I only needed to add that to the bootstrap script for installing policyrd-script-zg2, which is in universe. The bootstrap script removes 'universe' after that (that was already in the script, I'm guessing there is some other script I'm missing)
[00:11] <wgrant> pgquiles: Ah, I see.
[00:12] <pgquiles> wgrant: my modified bootstrap: http://www.elpauer.org/tmp/bootstrap
[00:13] <wgrant> pgquiles: You can probably give deboostrap an argument to do that for you.
[00:15] <pgquiles> mmm maybe --components=universe, I'll try tomorrow
[00:16] <pgquiles> now really off to bed :-)
[02:49]  * thumper needs JS help
[02:49] <thumper> rockstar: ping?
[02:49] <thumper> mars: ping?
[02:50] <thumper> ProtocolErrorException: Protocol error: 405
[02:50] <thumper> WTF?
[02:50] <thumper> anyone got any clues?
[02:50] <thumper> it is an XMLHTTPRequest
[02:51] <cody-somerville> Method not allowed
[02:52] <thumper> but why?
[02:52] <thumper> it is going from bugs.launchpad.dev -> code.launchpad.dev
[02:52] <thumper> is that allowed?
[02:53] <thumper> if not, I'm going to have to reconsider an approach
[02:53] <cody-somerville> did you use put or post?
[02:53] <thumper> GET
[02:53] <thumper> ah
[02:53] <thumper> no
[02:53] <thumper> WTF?
[02:53] <thumper> OPTIONS
[02:54] <thumper> why is it asking for options?
[02:54] <thumper> dumb yui
[02:55] <thumper> tracing NET with firebug causes firefox to hang for ages on each refresh for me
[02:55] <thumper> :(
[02:57] <ajmitch> bugs.launchpad.dev -> code.launchpad.dev might be considered cross-domain & forbidden, I'm not sure on the rules for how that is set
[02:58] <cody-somerville> yui should provide a way around that
[02:58]  * thumper sighs
[02:58] <wgrant> That counts as cross-domain.
[02:59] <wgrant> You cannot work around it.
[02:59] <wgrant> But why are you wanting to do that? The API is the same on all vhosts.
[02:59] <thumper> wgrant: I was just fetching an URL
[03:00] <thumper> wgrant: obviously I need to change that
[03:00] <thumper> :(
[03:00] <thumper> clucking bell
[03:00] <wgrant> What are you trying to do?
[03:00] <thumper> wgrant: show more details about the merge proposal on the bugs page
[03:00] <thumper> which includes a link to the diff
[03:01] <thumper> which will popup on the bugs page
[03:01] <thumper> got it working on branches
[03:01] <wgrant> Ah, I see.
[03:01] <thumper> it just uses Y.io
[03:01] <thumper> but doesn't work cross-domain
[03:02] <thumper> also
[03:02] <thumper> I really wanted some custom tags
[03:02] <thumper> but that is HTML5
[03:02] <thumper> well, custom attributes that is
[03:02] <thumper> beuno: I can haz HTML5 plz
[03:03] <cody-somerville> thumper, http://developer.yahoo.com/javascript/howto-proxy.html
[03:04] <cody-somerville> cross-domain is easy to work around using a server side proxy
[03:04] <wgrant> But that is crazy.
[03:04] <cody-somerville> no its not
[03:04] <cody-somerville> its the accepted best practice
[03:04] <thumper> cody-somerville: I think the API is the way to go in this case
[03:04] <cody-somerville> almost all javascript libraries come with support for it
[03:05] <wgrant> In this case the target is controlled by thumper too.
[03:05] <wgrant> So proxying it is just foul and ugly.
[03:07]  * cody-somerville agrees.
[03:08] <cody-somerville> wgrant, are you using ec2 btw?
[03:08]  * cody-somerville is in the process of signing up.
[03:11] <wgrant> cody-somerville: I've only used it a few times.
[03:11] <wgrant> Stuffing around + a couple of ec2tests.
[03:13] <ajmitch> how much does it cost to run through the test suite?
[03:13] <wgrant> A couple of dollars, I'd say.
[03:14] <cody-somerville> really?
[03:14] <cody-somerville> how long does it take to run?
[03:14] <wgrant> 3 or 4 hours on a high-CPU instance.
[03:14] <wgrant> IIRC.
[03:14] <wgrant> Or was it 2.5 hours.
[03:14] <wgrant> I don't remember.
[03:25] <thumper> can anyone recommend a good recursive diff?
[03:25] <thumper> for comparing trees
[03:25] <thumper> nm
[03:25] <wgrant> diff -r?
[03:27] <cody-somerville> hmm... I get ImportError: No module named _pythonpath when I try to run ec2 test
[03:27] <thumper> damn
[03:27] <thumper> must be friday
[03:28] <ajmitch> friday & nearly beer o'clock
[03:28] <cody-somerville> Its only Thursday here :(
[03:32]  * thumper wants to put his fist through the laptop right now
[03:56] <thumper> how do you get an attribute of an object using the xhtml representation?
[03:59] <wgrant> I doubt that you can; the XHTML representation is meant to be customised.
[04:04] <thumper> wgrant: yeah, I'm emailing the dev list
[04:04] <thumper> I think it is possible, I just don't know how
[04:07] <thumper> email sent
[04:07] <thumper> beer o'clock
[04:09] <ajmitch> the magic words have been spoken
=== cody-somerville_ is now known as cody-somerville
[04:50] <jml> hey
[04:50] <jml> my patch got rejected because we are in testfix mode
[04:50] <jml> wtf
=== geser_ is now known as geser
=== mwhudson_ is now known as mwhudson
[08:58] <adeuring> good morning
[09:10] <mwhudson> bigjools: gee thanks :)
[09:10] <bigjools> mwhudson: don't mention it :)
[09:10]  * bigjools notices it's Friday 13th
[09:27] <poolie> hello jml!
[09:27] <poolie> i've already run rocketfuel-setup on this machine
[09:27] <jml> poolie, oh
[09:27] <poolie> so i don't know if this is a good test
[09:28] <jml> poolie, hmm.
[09:28] <jml> poolie, thanks.
[09:28]  * jml has to re-do the whole thing anyway
[09:35] <wgrant> jml: Which technique did you end up going with?
[09:41] <jml> wgrant, a tarball with the repo, sourcecode and download-cache
[09:41] <jml> wgrant, the downside is that it requires 250MB-300MB of package data.
[09:42] <wgrant> jml: Either tar up /var/cache/apt/archives, or call apt-get with '-o Apt::Install-Recommends=no'
[09:42] <wgrant> jml: By excluding Recommends, you omit TeX and lots of other big stuff.
[09:43] <jml> wgrant, if I did archives, people would just copy stuff off the USB stick to their archives/ dir?
[09:43] <wgrant> jml: Correct. apt will see it there and use it.
[09:44] <jml> wgrant, will the relevant packages still be on my machine?
[09:44] <jml> I installed LP an awfully long time ago
[09:44] <wgrant> jml: Probably not. Best to do it in a chroot, anyway.
[09:48] <jml> or maybe I'll just rely on there being a mirror at UDS
[09:48] <jml> hmm.
[09:48] <wgrant> Possibly.
[09:48] <wgrant> But who knows how bad it will be this time.
[09:49] <jml> I admire your cynicism.
[09:49] <jml> the thing is I'll need 64 & 32 bit packages anyway.
[09:50] <mwhudson> start at http://archive.ubuntu.com and get clicking
[09:51] <jml> :(
[09:51] <wgrant> jml: Without recommends, it needs only ~100MB.
[09:51] <mwhudson> wgrant: how likely do you think it is we actually depend on a recommend ?
[09:51] <jml> wgrant, ok. I'll patch rocketfuel-setup to do that
[09:52] <wgrant> mwhudson: Not at all. I run without them.
[09:52] <mwhudson> wgrant: oh good
[09:54] <wgrant> http://paste.ubuntu.com/317645/ is the list of skipped packages. Only questionable things there are pyflakes and python-docutils, and the former is definitely optional.
[09:54] <wgrant> And I'm not sure the latter is used at all.
[09:55] <mwhudson> pyflakes is used by 'make lint' i guess
[09:55] <mwhudson> docutils, hm
[09:56]  * wgrant removes it and tries.
[09:58] <mwhudson> it's in versions.cfg
[09:58] <mwhudson> though heavens knows why
[09:58] <wgrant> Ah, yes, it's an egg.
[09:58] <mwhudson> and scripts/_ginalog.py but well
[09:58] <wgrant> So we don't need the package.
[09:58] <mwhudson> oh right yes
[09:59] <mwhudson> we also don't need the egg i would guess
[09:59] <wgrant> WTF is ginalog!?
[09:59]  * wgrant greps.
[10:00] <mwhudson> oh right docutils is used in generating the apidocs
[10:02] <jml> which is obviously a critical thing to build in every single branch
[10:07] <mwhudson> clearly
[10:15] <pgquiles> I am supposed to create my own production configuration, disabling devmode, etc, right? or is there a make run_all_production I'm not seeing?
[10:49] <jml> hello
[10:49] <jml> I have fixed the test failure.
[10:59] <stub> jml: It hasn't got to buildbot yet
[11:00] <wgrant> Does a testfix immediately revert the testfix state, or must buildbot run the test suite first?
[11:00]  * jml doesn't know
[11:01] <mwhudson> something in between
[11:01] <stub> I believe it reverts immediately
[11:01] <mwhudson> it reverts when buildbot-poll sees the change
[11:01] <mwhudson> and it runs out of a */5 cron i think
[11:01] <mwhudson> this is explained https://dev.launchpad.net/Trunk/Glue
[11:02] <stub> Ahh... its still with pqm
=== danilo-afk is now known as danilos
=== matsubara-afk is now known as matsubara
[11:37] <mrevell> hi
[12:22] <sinzui> danilos: ping
[12:23] <danilos> sinzui, hi
[12:23] <sinzui> danilos: I am looking at https://lpbuildbot.canonical.com/builders/db_lp/builds/244/steps/shell_7/logs/summary I cannot reproduce this failure on my system or in ec2.
[12:23] <sinzui> danilos: The failing line was
[12:23] <sinzui>         self.assertEqual(
[12:23] <sinzui>             self._getReviewables(no_older_than=self.base_time), [self.pofile]
[12:24] <danilos> sinzui, let me take a look in the test
[12:24] <sinzui> danilos: ^ What do you think, Could this be a time issue in the test suite?
[12:25] <danilos> sinzui, it could be, but I believe the times used in the test are in days, so highly unlikely, let me check
[12:31] <sinzui> danilos: I am inclined to force a build. I think the failure is spurious. I am not sure in this case though if I should disable the test. I wonder if the issue was really somewhere else in the suite.
[12:31] <danilos> sinzui, actually, it doesn't, it uses one millisecond, but I don't see how/why this should fail anyway... it writes the dates directly as two values where one is base_time + timedelta(0,0,1)
[12:32] <sinzui> Can I change it to an hour?
[12:36] <danilos> sinzui, sure
[12:36] <sinzui> danilos: I changed the test to an hour
[12:36] <sinzui>     later_time = self.base_time + timedelta(0, 3600)
[12:36] <sinzui> and it still passes. I want to land this change
[12:37] <danilos> sinzui, I suspect this is more likely related to commits than time resolution, but if you see further problems with it, please file a bug and I'll look into it shortly
[12:38] <sinzui> I doubt there will be further problems.
[13:28] <mwhudson> allenap: i will reply to your your reply to my mail
[13:28] <mwhudson> allenap: but not for a little while :/
[13:29] <allenap> mwhudson: Thanks, I look forward to it :)
=== matsubara is now known as matsubara-lunch
[15:07] <mars> mrevell, ping
[15:08] <mrevell> hi mars
[15:08] <mars> hi mrevell, just looking at the styling issue
[15:08] <mrevell> ah yes
[15:08] <mars> here, I'll send you a screenshot from my system
[15:13] <mars> BjornT_, ping, looking at mrevell's popup help styling issue.  Looks like the YUI folder is not being linked in devel?
[15:13] <mars> BjornT_, visit https://edge.launchpad.net/+help/openpgp-keys.html, check the CSS stylesheets - they are returning 404s
[15:14] <mars> so, what landed on devel to cause that...
[15:17] <mars> BjornT_, I suspect it broke recently: my yui-3final-upgrade branch is styled properly, so either a) it broke last night, since I pulled from trunk, or b) it broke in devel a while ago, and my branch fixes it
=== barry is now known as barry_
=== barry_ is now known as barry
[15:40] <pgquiles> maxb: wrt chroots, more or less answered in #debian-devel
=== matsubara-lunch is now known as matsubara
[15:42] <mars> BjornT_, looks like the CSS URLs were broken before the sprint.  Your build system changes maybe?
[16:20] <BjornT_> mars: maybe. could be that the 'current' symlink didn't get included with my changes
[16:20] <BjornT_> mars: a quick fix would be to change the css link to point to 3.0.0pr2 instead of current, which probably makes sense anyway
[16:21] <BjornT_> mars: this url works: https://edge.launchpad.net/+icing/yui/3.0.0pr2/build/cssreset/reset.css
[16:22] <mars> BjornT_, ok.  I'll see if I can identify where the CSS URL is generated
[16:22] <mars> and change it
[16:38] <BjornT_> mars: hmm, might be some other problem. i merged in the latest yui-3final-upgrade and lazr-js trunk into my branches, and i get similar errors for js files
[16:39] <mars> BjornT_, oh?  I tried my branch here, and the style links were fine. Could be an earlier build artifact though
[16:40] <BjornT_> mars: could be a lazr-js change then. are you using latest trunk?
[16:40] <mars> BjornT_, I tried it with the latest trunk, and it was broken.  I also tried it with trunk/ from last Friday, still broken.
[16:42] <BjornT_> mars: oh wait, looks like i'm using an old egg, and not my dev egg...
[16:43] <BjornT_> mars: that's a bug in the build system. bin/jsbuild doesn't update symlinks. it only checks whether there is a symlink, not that it's pointing to the right location
[16:44] <mars> BjornT_, should 'make clean' or 'make clean_js' handle that then?
[16:45] <BjornT_> mars: wel, make clean should remove the build directory, but bin/jsbuild should be able to update an existing build dir
[16:46] <mars> BjornT_, check the upgrade diff, line 40: http://pastebin.ubuntu.com/317110/
[16:46] <mars> BjornT_, that could be extended.  'make clean' calls 'make clean_js' as well.
[16:48] <mars> I actually wish that target was called 'jsclean', not 'clean_js', but we already have a 'clean_code' target in there.
[16:49] <BjornT_> mars: right, the Makefile changes look good
[16:50] <BjornT_> mars: bin/jsbuild should still be fixed, though, since people usually doesn't do 'make clean'
[16:51] <mars> BjornT_, good point
[16:51] <mars> you know, if we had a 'make update' target that did run 'make clean' for you... :D
[16:53] <BjornT_> mars: i only run 'make' and 'make run', and i wouldn't want any of them to automatically run 'make clean' :)
[16:54] <mars> BjornT_, why not?  If 'make clean' does the right thing, then it should only clean up generated build artifacts.  Nothing you hand-edited should be affected.
[16:57] <BjornT_> mars: run. but i see you don't build lp often :) it takes a long time building after a 'make clean'...
[16:57] <mars> that is true
[17:08] <Ursinha> hi sinzui, I'm trying to go to https://code.edge.launchpad.net/startup-manager/trunk/+addbranch and am getting a similar oops as the one when trying to subscribe to bugmail
[17:10] <Ursinha> sinzui, is the problem related? https://lp-oops.canonical.com/oops.py/?oopsid=OOPS-1410G2138
[17:10] <sinzui> I am not certain
[17:11] <sinzui> Ursinha: I am logged in so I should not see the message
[17:11] <sinzui> Ursinha: how did you get to the link? I do not see it
[17:12] <Ursinha> sinzui, https://edge.launchpad.net/startup-manager/trunk, click on Submit code
[17:12] <sinzui> Launchpad is crack for showing overview pages on the code hsot
[17:12] <sinzui> ah, from the involvement menu
[17:12] <sinzui> That is very difference
[17:13] <sinzui> The error is true. you submit code to a project, not a series.
[17:13] <sinzui> That is why we are seeing a 404
[17:15] <Ursinha> sinzui, bug 482256
[17:15] <mup> Bug #482256: 'Submit code' OOPSes in the project series overview page <oops> <Launchpad Registry:New> <https://launchpad.net/bugs/482256>
=== barry` is now known as barry
[18:01] <mrevell> Okay, night all. See you in Dallas if you're gonna be there.
[19:17] <Ursinha> hey rockstar, are you there?
[19:53] <rockstar> Ursinha, hi
[20:00] <rockstar> Ursinha, what's up?
[20:00] <Ursinha> rockstar, I'm trying to remember :)
[20:00] <rockstar> Ursinha, okay.  Well, we're back from lunch, but deryck and I are going to pair for a bit, so I may have to ignore you.
=== matsubara is now known as matsubara-afk
[20:27] <EdwinGrubbs2> mars: ping
=== EdwinGrubbs2 is now known as EdwinGrubbs
[20:28] <mars> hi EdwinGrubbs2
[20:28] <thumper> rockstar: you around?
[20:28] <rockstar> thumper, I am.
[20:29] <thumper> rockstar: I emailed my javascript frustrations to launchpad-dev
[20:29] <rockstar> thumper, yes, I read it.
[20:29]  * mars has not gotten to it yet :(
[20:29] <thumper> rockstar: intellectronica answered some, but didn't really tell me much new (except the hidden <a> which I'll probably use)
[20:29] <thumper> rockstar: 2 remaining items
[20:30] <intellectronica> thumper: what else is still unclear?
[20:30] <EdwinGrubbs> mars: I've been working on getting some lazr-js tests pass in IE. hasAttribute('tabIndex') always returns true in IE, but getAttribute('tabIndex') returns 32768, so I'm thinking of changing the test to pass if one of those is correct. Do you have any better ideas?
[20:30] <thumper> rockstar: *how* to I use LP lib to get the preview_diff attribute of the MP as xhtml?
[20:30] <thumper> rockstar, intellectronica: and why does the diff look different in the popup vs in-page
[20:30] <rockstar> thumper, intellectronica can probably help there.  I don't think that interface is flexible enough for what I need, so I haven't used it.
[20:31] <salgado> thumper, accept: LP.client.XHTML in the config
[20:31] <intellectronica> thumper: when you make a request using the lp client, pass `accept: LP.client.XHTML` as part of the configuration object
[20:31] <rockstar> thumper, flacoste has drafted a new proposal that's less restful but more practical, which would then help.
[20:31] <thumper> intellectronica: is there an example of attribute access using xhtml in the bugs JS somewhere?
[20:31] <rockstar> thumper, wait, why do you want to get the preview_diff HTML through the API?
[20:32] <intellectronica> thumper: yes, look at bugtask-index.js, for example
[20:32] <thumper> rockstar: because you can't it the code.lp.net from the bugs.lp.net page
[20:32] <thumper> rockstar: it won't let you
[20:32] <thumper> I don't know why
[20:32] <thumper> it is not posting, just getting
[20:32] <mars> EdwinGrubbs, nope, sounds like a fun IE bug.  Have you considered using .get('tabIndex')?  They may have normalized the return value in the YUI code
[20:32] <rockstar> thumper, bugger.
[20:32] <thumper> rockstar: yes
[20:32] <thumper> ok,
[20:33] <thumper> that just leaves the last point
[20:33] <thumper> I have the same code generating the diff for the in-page view and the popup
[20:33] <thumper> but when I set the content of the pretty overlay
[20:33] <thumper> it renders the table outside the div it should be in for some freaking weird reason
[20:33] <EdwinGrubbs> mars: get('tabIndex') is way worse. It returns 0. The difference may be due to one using the IE extension to getAttribute and the other not. http://msdn.microsoft.com/en-us/library/ms536429%28VS.85%29.aspx
[20:34] <mars> EdwinGrubbs, isn't '0' what you want?
[20:34] <mars> since that is the default?
[20:34] <rockstar> thumper, that might have to do with markup and css.
[20:34] <intellectronica> thumper: actually, an example more relevant to your case is get_comment_HTML in lib/canonical/launchpad/javascript/lp/comment.js
[20:34] <thumper> :(
[20:34] <thumper> intellectronica: thanks
[20:34] <EdwinGrubbs> mars: well, on firefox get('tabIndex') returns -1, so it's not consistent.
[20:35] <rockstar> thumper, so you might have to mess with the markup it creates.
[20:35] <thumper> intellectronica: I have written an adapter already, needed for the commit message
[20:35] <thumper> rockstar: any ideas what would cause it to render outside?
[20:35] <rockstar> thumper, honestly, I expected that when you were talking about it.  It may need a separate view entirely.
[20:35] <thumper> rockstar: it isn't just rendering, the firebug dom shows it outside
[20:35] <rockstar> thumper, not without looking at the exact markup.  It might be using floats or tables or any number of bad things.
[20:35] <thumper> rockstar: there are tables
[20:36] <mars> EdwinGrubbs, some raw data: http://www.weba11y.com/blog/2007/11/
[20:36] <thumper> but the table is in a div, so should be ok
[20:36] <rockstar> thumper, yea, tables are amazingly complex.
[20:36] <rockstar> thumper, I probably can't look now, but if you point me to a branch, I'll do my best to get to it.
[20:36] <thumper> the fmt:diff renders a table
[20:36] <thumper> rockstar: my popup-diff branch
[20:37] <thumper> rockstar: and thanks
[20:37] <rockstar> thumper, ah, okay.  It was in my queue of things to do today, but deryck and I had yaks to shave.
[20:37] <thumper> I'll be working on this later, as I'm using it as part of my assignment due Monday:)
[20:37] <mars> EdwinGrubbs, try: node.get('tabindex');  // note the lowercase 'i'
[20:37] <thumper> rockstar: I did change the behaviour so it gets a spinner after the link, nice fallback to the librarian file, and popup only once loaded though
[20:38] <thumper> rockstar: it was pretty easy to do
[20:38] <EdwinGrubbs> mars: from that blog post, it seems like you would want NoTabIndex to set the tabIndex to -1 instead of using removeAttribute('tabIndex')
[20:38] <thumper> rockstar: took maybe 10 minutes
[20:38] <rockstar> thumper, great.
[20:38] <thumper> I'm off to buy meat for the BBQ :)
[20:39] <EdwinGrubbs> mars: .get('tabindex') is a separate variable than getAttribute('tabindex') since get() is case-sensitive.
[20:39] <mars> EdwinGrubbs, yes, it looks like you are right
[20:40] <rockstar> EdwinGrubbs, we might want to document that somewhere.
[20:41] <mars> EdwinGrubbs, there is also this: http://developer.yahoo.com/yui/3/node-focusmanager/index.html
[20:41] <EdwinGrubbs> rockstar: there will definitely be a very verbose comment about this.
[20:43] <mars> EdwinGrubbs, trying to remember if setting tabIndex to -1 on the overlay <div> will remove all of its child controls from the tab order as well...
[20:44] <mars> EdwinGrubbs, according to http://www.w3.org/TR/html401/interact/forms.html#adef-tabindex, the value needs to be between 0 and 32767.  So IE is setting it to the last possible tab element on the page.
[20:45] <EdwinGrubbs> mars: it's actually setting it to 32768, which is an illegal value.
[20:45] <mars> ah
[20:47] <mars> EdwinGrubbs, eewww: On IE it is not possible to determine whether a div has no tabindex or
[20:47] <mars> tabindex="0" using elem.getAttribute("tabindex") or elem.tabIndex
[20:47] <mars> because a default value of 0 is returned. Microsoft provides an
[20:47] <mars> extension to getAttribute() that can be used to determine if the
[20:47] <mars> tabindex is unset. If a second parameter of value "2" is passed to
[20:47] <mars> getAttribute(), a value of 32768 is returned whenever tabindex is not
[20:47] <mars> set. See Microsoft's documentation on getAttribute:
[20:47] <mars> double-eewww
[20:47] <EdwinGrubbs> mars: yes, that's the link I sent you
[20:48] <mars> EdwinGrubbs, yep, but I didn't understand why it was necessary.  That is not fun.
[20:50] <EdwinGrubbs> mars: so my main question is whether I should just fix the unit test, or if I should also have NoTabIndex set it to -1.
[20:51] <mars> EdwinGrubbs, set it to -1.  That is a modern browser extension.  Don't worry about IE.  Please ensure that child elements in the Overlay are still tab-navigable.
[20:52] <mars> EdwinGrubbs, if IE is being a pain, then put an exception into the test just for it, with the relevant links.  I don't see a better way to handle this at this time.
[21:02] <EdwinGrubbs> mars: IE seems to handle -1 exactly like the other browsers. It was just the removeAttribute() that was really difficult to verify, and I didn't know if there was a reason NoTabIndex should use that. I will test the overlay, and I think -1 is much better solution for NoTabIndex.
[21:02] <mars> EdwinGrubbs, should the extension be renamed to RemoveTabIndex then?
[21:03] <mars> or RemoveFromTabOrder, or something
[21:09] <EdwinGrubbs> mars: hmmm, maybe DisabledTabIndex.
[21:09] <mars> EdwinGrubbs, DisableTabIndex sounds good.  Y.augment(Foo, DisableTabIndex);
[21:10] <EdwinGrubbs> ok
[21:10] <EdwinGrubbs> mars: can I send you the reviw later today, so you can look at it Monday? I have some other IE fixe in the branch, but it's not that big.
[21:10] <mars> EdwinGrubbs, sure
[21:13] <wgrant> pgquiles: Why are you now installing debhelper and python-twisted?
[21:13] <wgrant> Neither of those are necessary.
[21:14] <wgrant> pgquiles: And make run_all doesn't do anything extra that Soyuz needs.
[21:37] <pgquiles> wgrant: launchpad-buildd requires debootstrap and python-twisted, IIRC
[21:37] <pgquiles> wait, debhelper?
[21:38] <wgrant> pgquiles: launchpad-buildd does need python-twisted, but it is included in the package's dependencies. It does not require debootstrap.
[21:39] <wgrant> And you only need debhelper if you are building packages yourself.
[21:39] <pgquiles> wgrant: yeah, that was it. But the next step in the howto is building the launchpad-buildd package, so I need debhelper.
[21:40] <wgrant> pgquiles: Ohh, true.
[21:40] <wgrant> But debootstrap/python-twisted are not required.
[21:41] <pgquiles> dpkg -i launchpad-buildd fails, debootstrap and python-twisted are dependencies in debian/control
[21:41] <pgquiles> (IIRC, I don't have a checkout here)
[21:42] <wgrant> pgquiles: Ah, then perhaps you meant 'apt-get -f install'
[21:43] <wgrant> No point installing them explicitly, since those deps could change.
[21:43] <pgquiles> wgrant: mmm when I install launchpad-buildd with dpkg -i as said in the howto, dpkg returns an error
[21:44] <pgquiles> I'm trying to browse code.launchpad.net to find the control file and check the dependencies for launchpad-buildd
[21:44] <wgrant> pgquiles: If you then apt-get -f install, it should finish installing the package.
[21:44] <pgquiles> ah, yes
[21:44] <wgrant> Depends: python-twisted, debootstrap, dpkg-dev, linux32, file, bzip2, sudo, ntpdate, adduser
[21:44] <pgquiles> yeah
[21:44] <wgrant> It should not depend on debootstrap, but that's another matter.
[21:45] <pgquiles> after my research yesterday and a bit of IRC talking with lamont this today, I think I know how to build a chroot like the ones you can download form launchpadlibrarian.net
[21:46] <pgquiles> I need to find out about some config files but I'll probably write an article about that this weekend
[21:46] <wgrant> pgquiles: What additional stuff was necessary?
[21:47] <pgquiles> wgrant: it's esentially debootstrap --variant=buildd --components=main,universe, then chroot to that folder, create buildd group and user and run root/bootstrap
[21:47] <pgquiles> dump kernel config and two or three more config files I can't remember now