[05:40] wgrant: ping? [05:42] NCommander: Hi [05:42] wgrant: so I was trying to use the raw-installer custom upload to publish squashfs into installer-arch into the dist folder [05:42] Obviously it didn't work, and since you've noticed it, I suspect I broke something else in LP [05:43] Yaeh [05:43] If you're going to commit such evil, perhaps give us a headsup first :) [05:43] raw-installer is barely tested to work in PPAs at all, let alone at that scale. [05:44] * NCommander gulps [05:44] What did I hose >.>; [05:45] (raw-installer generally works with d-i, guess I broke the world somewhat more) [05:47] NCommander: https://oops.canonical.com/oops.py/?oopsid=OOPS-8a6501691b7c6ada250ac4ab32b1ee69 [05:47] https://lp-oops.canonical.com/oops.py/?oopsid=8a6501691b7c6ada250ac4ab32b1ee69 [05:48] NCommander: I'm not quite sure why it's crashing like that, but it's probably something evil in your tarball [05:48] wgrant: oh, that was the first one [05:49] The code is hardcoded to attempt to try and extract installer-$ARCH [05:49] FIrst attempt had livecd-$ARCH, and I thought it merely published nothing [05:49] THe second attempt had the folder right, and then nothing showed up so I ran away quickly :-) [05:51] * NCommander decides it *might* be easier to setup a local LP instance to test and abuse this less I bring the wrath of the Soyuz devs upon me [05:51] NCommander: All three Soyuz devs share a timezone now, too .... [05:51] but fortunately, there an ocean away from me; gives me plenty of time to hide :-) [05:51] =:o [05:51] *their [05:53] Might be easier to simply implement a new custom upload type but I think it might cause someone to maul me [05:58] NCommander: Still in AKL? [05:58] Migrated back to OR after a check engine light and other concerns forced me to call it quits [05:58] * NCommander did accomplish everything on the TODO list except the Dempster Highway [05:59] more annoyingly, the error codes cleared themselves once I was about halfway back to Oregon, but I have that nasty feeling I'm looking at transmission failure sooner rather than later [06:01] * StevenK rejects NCommander's uploads with prejudice. [06:02] NCommander: What do you drive? [06:02] * NCommander hasn't uploaded anything to the archive in awhile [06:02] StevenK: 2007 Hyundai Tucson [06:02] 2WD, but it gets excellent mileage and is decent on washboard roads (and arctic tundra) [06:03] NCommander: http://en.wikipedia.org/wiki/Subaru_Legacy_%28fifth_generation%29 is what Sarah and I drive. [06:03] http://en.wikipedia.org/wiki/Hyundai_Tucson - me [06:03] Yeah, I found that already [06:03] Hilariously, in NYC, when I park the damn thing, I get the compact discount, and the SUV surcharge [06:03] NCommander: Your PPA uploads are what I rejected. [06:04] Wait, what? [06:04] What queue did they end up in? [06:04] ACCEPTED [06:04] ... [06:04] I *really* broke LP [06:04] PPA uploads are automatically accepted [06:04] And crashing doesn't autoreject [06:05] Since it usually indicates an LP bug. [06:05] Right [06:05] So LP got stuck in an infinite loop? [06:05] It will try and process the queue, hit your upload, crash and die. [06:05] It tried to reprocess them a couple of hundred times yesterday [06:05] Just makes things a bit slower. [06:06] Oh, that's right, it will just skip your PPA [06:06] NCommander: Auto or manual? [06:06] auto unfortunately [06:06] Hah [06:06] Parents wouldn't co-sign the loan unless it was something they could drive [06:06] NCommander: Said Legacy (which is sold in .au as 'Liberty') is 6 speed manual 4WD [06:07] Bah [06:07] Damn you and your stickshift [06:07] Wait. [06:07] NCommander: If you tried to drive it, you'd end up banging your hand into the door handle from muscle memory [06:07] You took and automatic to Alaska? [06:08] ScottK: semi-automatic [06:08] ScottK: He's brave and/or stupid. [06:08] P-R-N-D-1-2-3-4 [06:08] (with 1-4 being sequential) [06:08] so I can gear down/up like a stick but no clutch :-/ [06:08] That's an automatic. [06:09] http://goo.gl/maps/m2YDl - (route I drove going up) [06:10] "This route crosses through Canada." [06:10] Hahaa [06:10] You pretty much have no choice, though? [06:10] It's that or a boat. [06:10] Right [06:11] No so much fun driving on a boat [06:11] that's a long way [06:11] Depends on how big it is and how many people are walking around on the vehicle deck. [06:12] THere is the Alaska Martime System from Bellingham, WA to southern Alaska [06:12] But the prices make an international business class fair look cheap [06:17] wgrant: honest opinion, if I wanted to build livefs's on the buildds would you say the 'raw-installer' custom upload is approiate, or I should create a new one (or some pre-existing one somewhere)? (I think raw-installer does some managing of the "current" directory in installer-*) [06:18] NCommander: I'd create a new one. Do they even need to be published to the archive? [06:18] wgrant: I was thinking a PPA would do the trick [06:19] NCommander: I'd go with something like translations-static, where you pull the files down through the API [06:20] wgrant: infinity was working on modifying LP that buildd-* could accept a job from ubuntu-cdimage, but my mental thought was it isn't hard to generate it via PPA. I did need a little bit of voodoo though to get livecd.sh to run in a PPA [06:20] * NCommander was trying to figure out a proof of concept before I broke the world [06:21] Interesting concept of time you have. [06:21] "Before" [06:21] then when we need to respin a squashfs, nusaken can generate the stub source package, fire it into a PPA, and poke liblaunchpad to see "has it built yet?" then download [06:21] ScottK: NCommander saying he broke the world was a common occurance on the Ubuntu Mobile weekly calls. [06:22] You'd think he'd have learned by now. [06:22] * NCommander upgraded from breaking the world to merely breaking davidm [06:22] NCommander: ITYM nusakan [06:22] but, just for the record, you guys suck :-P [06:23] NCommander: We aren't the one that drove to Alaska with an automatic transmission. [06:24] eh, that's not the crazist thing I ever did [06:24] Color me not surprised. [06:24] * NCommander only notes now that he's digging himself deeper [06:24] NCommander: Dig up, stupid. [06:24] Or at the very least stop. [06:27] WIth friends like you, who the heck needs enemies :-P [06:28] * ScottK misread that last word at first and thougth things had taken a seriously weird turn. [06:30] * NCommander doesn't want to know what ScottK read [06:30] For once, you're right. [06:31] haha [06:31] I can guess, and I don't like it. [06:33] ok, how do I get vi in busybox [06:33] bigjools: its a compile time option [06:33] arse [06:33] bigjools: its possible its there but no symlink [06:33] doesn't help with my current busybox shell then! [06:34] try busybox vi [06:34] applet not found [06:34] I believe ed is standard in busybox though [06:34] not in this one! [06:34] bigjools: what are you trying to fix? [06:34] * NCommander *might* have a statically linked busybox handy [06:35] I am booting off a usb stick so I can edit the password file a box [06:35] on a* [06:35] Oh [06:35] That's easier [06:35] chroot *mnt directory* /bin/bash [06:35] export TERM=vt100 [06:35] nano /etc/passwd [06:35] I wonder if in-target will work [06:35] bigjools: Why not just use GRUB to do that? [06:36] This isn't Windows :) [06:36] * NCommander rather do it from an EFI shell :-P [06:36] that assumes that I can get into grub [06:36] got away with no editor [06:37] is the file malformed or just forgotten password? [06:37] chroot did it, why did I not do that earlier when I thought of it ... [06:37] it's a maas-installed juju node where something went wrong in cloud-init [06:38] thx for tip NCommander [06:39] * NCommander would hope that a juju wouldn't bring pain and destruction to innocent installation [06:39] ^s [06:39] Interesting theory. [06:39] well it's supposed to add an ssh key... but failed [06:40] .... How do you go from add an ssh key to OMG BROKEN /etc/passwd O_O; [06:40] it's not broken [06:40] Why do you want to upload the squashfs to an archive instead of to the librarian? [06:40] I am addingt a password [06:40] so I can ssh in [06:40] * cody-somerville is still baffled that you guys want to use launchpad buildds instead of using openstack cloud + juju or whatever. [06:41] cody-somerville: mostly because I didn't consider uploading to librarian. [06:41] cody-somerville: generally speaking, I rather used tried and tested tools for critical infrastructure until they've had some time to bake :-) [06:42] Also, openstack is very x86 only ATM [06:43] I'm not sure how the new code you're writing to add support for this in Launchpad is tried and tested vs. OpenStack [06:44] and IIUC, Linaro 'cross-builds' their ARM images on x86 (but I might be mistaken on that) [06:44] Plus, there is also MAAS. [06:44] cody-somerville: only new code I'm writing is a publishing method into Launchpad. In addition, I don't want to spawn multiple sets of infrastructure we have to maintain [06:45] lol. Tell me about it. If you guys do implement this, I'm going to have to add support to Offspring to be able to use the launchpad buildd farm [06:45] not looking forward to that [06:45] As it stands we have to maintain a seperate set of buildds for live images, and another (much larger) set for package builds. OpenStack doesn't really solve the problem I want to solve [06:47] That's a fair point. [06:47] In addition, then the machiens used for liveFS builds can migrate into the builder pool, and add more ARM and PowerPC hardware [06:49] of course it helps if I didn't try and ssh into the BMC [06:54] cody-somerville: actually, I don't want to publish to librarian. Being able to simply wget squashfs's makes life easier when I'm tinkering with a local CD image build vs. having to deal with launchpadlib [06:55] Files in the librarian are wgetable [06:55] * NCommander has found that while its generally trivial to put something in librarian, it was my experience that getting things out are more difficult [06:55] cody-somerville: yes, on some impossible-to-remember URL which changes each time a new file is uploaded [06:56] you don't want a unique filename for each squashfs build you do? [06:56] We don't currently do it. SUppose if I'm already burning the existing livecd code to the ground, might as well implement it in a slightly smarter manner [06:57] other thing I need to figure out if is there is a saner way to actually get the correct privelleges to call livecd.sh (it doesn't like fakeroot :-)) [06:59] NCommander: If you have the LFA, it's pretty easy. :-) [07:01] LFA? [07:01] LibraryFileAlias [07:01] er [07:02] I meant on the buildds themselves [07:02] * NCommander was not happy with his workaround to make livecd.sh go [07:02] s/was/is/g [09:19] hiya [09:19] I was wondering if anyone was looking into https://bugs.launchpad.net/launchpad/+bug/1031764 right now [09:19] Ubuntu bug 1031764 in Launchpad itself "timeout on code.launchpad.net/~ubuntu-branches" [Critical,Triaged] [09:25] dholbach: Bug #1029642 is probably the one you actually care about [09:25] Launchpad bug 1029642 in Launchpad itself "ScopedCollection:CollectionResource:#branch_merge_proposal-page-resource (landing candidates) dying from late evaluation of security rules" [Critical,Triaged] https://launchpad.net/bugs/1029642 [09:25] aha! [09:26] wgrant, so is this a problem somewhere else as well? or just in the ~ubuntu-branches MP list? [09:26] dholbach: It's slow globally, but has only significant affected ~ubuntu-branches and a U1 thing, from what I've seen. [09:27] ok, that's good to know - thanks a bunch [09:27] It will hopefully be fixed soon. [09:29] awesome [09:29] thanks wgrant === Quintasan_ is now known as Quintasan [12:49] weird issue with packages copied from one ppa to another: the target ppa was empty to start with and after the copy there is no Release.gpg in the target ppa [12:49] known issue ? Should I file a bug ? === zyga is now known as zya-afk === deryck is now known as deryck[lunch] [16:45] I am trying to add a .desktop file to an existing package (that I didn't originally make). It uses cdbs in the rules file. I tried just putting blah.desktop in the debian/ directory but that wasn't enough. How can I make sure the .desktop file gets installed with the package? [17:52] I know this is an odd question but... when creating a recipe, can I omit the base branch and _just_ have two nested branches? [17:52] Is that even possible? [17:52] george_e: you can't nest a branch if you don't have something to nest it into [17:53] why would you want to do that anyway? [17:54] Well... I have one nested branch for the Debian packaging. [17:54] ...and I was wondering if the code could be nested in a directory as well. [17:55] So the uploaded .tar.gz would look something like this: ./debian ./the_code [17:55] The Debian packaging would be aware that the code is in a separate directory. [17:56] I guess it does sound a little strange. I'll explain a little further: [17:56] So I've got some Debian packaging that generates two binary packages - and each needs a separate source tree. [18:03] no, that's really not how you should handle the packaging. the debian/ dir should be nested under the top level dir of the source tree [18:03] why are you building two different sources in the same packaging? [18:27] hi i am getting the following http://pastebin.com/p9AXugEq [18:28] config file http://pastebin.com/rrsEEmMC [18:48] haseeb: has nothing to do with launchpad or bzr really, but it is a general ssh issue. you need to fix the file permissions on ~/.ssh and ~/.ssh/config [18:50] and your config is wrong; you'd need "Host bazaar.launchpad.net" and you can omit the "HostName" under that. also you don't need configure ssh to configure your lp hostname for bzr; you should do "bzr lp-login" instead [18:51] ok [18:55] dobey, same error [18:57] you still need to fix the file permissions on them then [18:58] or just delete it if that's the only thing in the .ssh/config file [19:03] dobey: Sorry, I was away for a few minutes there. I'm not really building two source packages - since the source tree is identical in both cases (I'm nesting the same branch in two places). The reason I'm doing this is because I'm building the library for two architectures. [19:04] Note that this is NOT the same as specifying Architecture: Any in the debian/control file. [19:04] ...because I am not building Linux / Unix binaries. [19:04] what are you trying to do exactly? [19:04] I am using Mingw-w64 to build Windows binaries. [19:05] (Sorry if this really is confusing.) [19:05] you probably don't want to use launchpad recipes to do that [19:05] So my plan was to have the Debian directory and two identical source trees. [19:05] or debian packaging [19:05] One for 32-bit Windows and one for 64-bit. [19:05] dobey: I do - because the files are used for cross-compiling. [19:06] I'm not building _executables_. [19:06] I'm building the .lib / .a files that the executables use. [19:06] launchpad recipes are for building debian packages for use on Ubuntu via a PPA [19:06] Exactly. [19:06] That's what I'm doing. [19:06] Windows is not Ubuntu and doesn't use PPAs [19:07] The packages contain files that go in /usr/i686... and /usr/x86_64. [19:07] dobey: Other distros do this. Example: http://rpm.pbone.net/index.php3?stat=26&dist=84&size=46569&name=mingw64-zlib-static-1.2.5-10.fc17.noarch.rpm [19:07] Those packages just contain the files that Mingw-w64 produces. [19:08] I'm not trying to stuff Windows applications into a Debian package. [19:08] These packages are just the development files that are used to produce them. [19:09] and mingw-w64 builds win64 binaries on i386 ubuntu? [19:09] Yes. [19:09] That's why the packages produced are architecture-independent. [19:10] The Mingw-w64 compiler can produce 32 and 64-bit Windows executables from both a 32 and 64-bit Ubuntu installation. [19:10] well, you don't need multiple source trees and everything nested separately [19:11] you need a more complex deiban/rules file [19:11] What else can I do? [19:11] dobey: Ah, okay. [19:11] How would that work? [19:11] in much the same way that some C extensions to Python are built for multiple python versions, for example [19:11] Would you like to see what I have so far? [19:11] I have a branch on LP. [19:12] or make your packaging branch the main tree, and nest everything else [19:12] depending on how the thing you're building actually configures/builds itself [19:12] Currently, the debian/rules here only builds for 32-bit Windows - http://bazaar.launchpad.net/~george-edison55/win-cross/zlib-debian/files [19:13] dobey: That was something I had considered (making the Debian branch the main one). [19:17] i'm not sure exactly how to do what you want, but something like http://bazaar.launchpad.net/~ubuntuone-control-tower/dirspec/packaging-dailies/view/head:/rules (how python3 and python2.7 packages are built from same source), might work for you [19:18] dobey: Okay, thanks. I have done some packaging before for Python modules that generate both Py2k and Py3k packages. [19:18] ...so hopefully something like that will work. [19:19] Maybe the Makefile will let me specify a directory for object / binary files. [19:24] george_e: it doesn't matter really, since install for one should get called before building the other anyway [19:24] george_e: unless you want to fix it to allow building both at the same time [19:24] Ah, so then I just need to run 'make clean' between the two then? [19:25] I guess I need to read up more on how dh_build works. [19:25] well, dh_auto_clean should be getting run between each build, yes [19:25] so you can use override_dh_auto_clean to remove any extra files you need to remove, as is done in that py3k example i showed [19:26] Right. [19:26] One last question... I assume override_dh_auto_build will be called once for each package. [19:26] How do I know which one is getting built? [19:27] i'm not sure. i don't think dh provides any magic for it automatically, so you'll probably have to do some extra stuff to tell it which one to build [19:27] Thanks for helping me out. === Guest78961 is now known as dpb__ === dpb__ is now known as dpb___ [20:59] so. I can't create a account (I don't receive any emails, my mailserver isn't contacted), when using the 'forgot password' functionality I end up with a OOPS ID and something wrong. When trying to use the SSO support they are amazingly unhelpful. What does one do? [21:17] svuorela: SSO support is where you need to go; have you used their email support system or web form or IRC ? [21:21] lifeless: the web form [21:23] svuorela: I suggest you try #canonical-isd on this IRC server [21:23] svuorela: but first [21:23] what URL are you using that has the 'forgot password' functionality on it. [21:24] https://login.launchpad.net/+login [21:27] lifeless: when is a appropriate timezone for heading into #canonical-isd ? [21:27] now [21:27] svuorela: thats definitely an ISD managed url, sorry! === heroux_ is now known as heroux