[00:00] lool: please go ahead and fix mbr if you know how - it's been on the guilt-inducing section of my to-do list for a while [00:00] (and thanks) === nenolod is now known as moogle [01:00] heya [02:06] If there's an archive admin with a spare moment, I'd appreciate it if you'd binary promote kdeedu-kvtml-data to make all of kdeedu installable again ... [02:14] done [02:14] slangasek: Thanks. [02:15] slangasek, do you want to review sion? ;] [02:15] * ScottK thinks he has a clean KDE slat to hand back over to Riddell now that he's returned from vacation. [02:15] cody-somerville: no, that requires concentration and I have a headache [02:16] slangasek, I promise its nice and clean ;] [02:16] but alas, headaches suck [02:29] * cody-somerville decides to merge valgrind. [02:59] hmm... [02:59] it doesn't seem possible to create an intrepid chroot [02:59] gnupg: Depends: libcurl3-gnutls (>= 7.16.2-1) but it is not installed [03:00] cody-somerville: s/intrepid/jaunty/? [03:01] nope [03:01] intrepid [03:02] Erm. [03:03] o_o? [03:03] Something like that, yes... [03:03] Which pockets are enabled? [03:04] wgrant, chroot creation has no pockets enabled [03:04] debootstrap can't use them [05:24] cjwatson: oh I did push it [05:25] I only identified the minimal fix for i386 and amd64 so far, I'm not yet sure about the lpia one, but my 0.11 upload to my ppa also built on lpia, so it shouldn't be too hard [05:31] If I have two binary packages building out one source package, do I need conflict/replaces lines if I need to move a file from one binary package to another? [05:32] NCommander: Why wouldn't you? [05:32] You should probably have them Conflicts: other-pkg (<= version-where-you-moved-the-file) [05:32] thanks [05:32] * NCommander had a brainfart [05:32] I can't remember if they replace each other as well or not [05:39] NCommander: just replaces, not conflicts [05:40] NCommander: If you move a file from a to b in new version 1.2-3 let b replace a (<< 1.2-3) [05:40] If you move all files / replace a package completely, then you want a conflicts to allow the package to select the package for removal, or a dummy package [05:41] Oh [05:42] lool, Replaces: python-qt4-dev (<< 4.4.2-3) - so that will do the trick? [05:42] (the package version is -3, replacing -2 and -2ubuntu1) [05:42] NCommander: I'd put a ~ on the end for it if gets backported. [05:43] ScottK, Replaces: python-qt4-dev (<< 4.4.2-3~) [05:43] Right? [05:43] NCommander: Yes [05:43] Yep [05:43] THanks lool and ScottK [05:43] We might have this fixed before the weekend is over [05:43] * NCommander waves flag [05:47] Wave your flag AFTER it's fixed. [05:50] NCommander: You're going to get this uploaded to Debian and we sync it? [05:50] Depends how fast I can get a sponsor [05:50] If so, yes, if not, no [05:51] Its going into experimental so no one is going to freak about "OMG lenny frozen" [05:53] NCommander: What package is this? [05:54] python-qt4 [05:54] I'm a member of the team that owns it in Debian [05:54] Oh it's DPMT [05:54] bingo [05:54] I guess I can sponsor that then [05:54] I need to do a final test build and make sure my changes actually work, so it may be awhile [05:55] (its building now) [05:56] Is it committed? [05:56] Cause I don't see it in trunk [05:57] No [05:57] Because ATM its not quite working right :-) [06:12] Argh the reason mbr/lpia built in my ppa is because ppas are built on amd64 hosts in a lpia chroot [06:12] lool, sorry about that [06:12] this laptop is a little unstable when I'm running a KVM :-/ [06:13] mcasadevall: You moved pyqtconfig from -dev to the main package? [06:13] lool, committed [06:13] * mcasadevall blinks === mcasadevall is now known as NCommander [06:13] CIA bot is handy. [06:14] Ah, I'm pretty sure that's not right. [06:15] ScottK, and making the dev package arch any is worse since thats the only arch dependent files in it [06:15] Hmmm. [06:16] * NCommander kicks xchat [06:16] Making dev arch any is a bad idea, and I don't think making yet another package for two files is a great idea either ... [06:17] * ScottK does svn up and looks [06:20] NCommander: Your rules has spaces instead of tabs [06:21] around "echo yes is needed" [06:22] I don't parse that echo yes is needed line BTW [06:24] lool, I didn't add that [06:26] It's all three of the lines of that comment. [06:26] ScottK: And further lines below [06:27] lool, I'll fix them, I'm just noting the original committer is who added them [06:27] ack [06:28] So I'm not too hot on this arch specific stuff, it doesn't really make sense to me [06:30] lool, whats the problem? [06:30] The file I moved into python-qt4 can be different on platforms without a FPU or ARM [06:30] (it contains the "handle float qreal code" switch) [06:35] NCommander: Looking at this file, it looks quite arch specific, but the whole concept less so [06:36] NCommander: Sorry but can you remimd me how python-qt4 fits in our problem? It's a dep of sip? [06:37] It's a dependency of kde4bindings [06:37] And also controls sip's configuration file [06:37] er [06:37] controls sip's configuration [06:37] i.e., what options should be enabled [06:37] What I'm unsure about is whether it's ok to rely on arch specific differences for this [06:38] huh? [06:38] Currently kde4bindings in the archive properly parsers this, so once python-qt4 is fixed, we just need to retry the build [06:40] NCommander: At a high level the problem is that arch specific constraints are not taken into account by a code generation tool; this code generation tool may be run on any arch and should produce code which builds on all arches; if you agree with this, then surely you'll agree that the code generation tool shouldn't rely on arch specific changes [06:40] lool, ? [06:41] What part do you "?" on? [06:42] Most/all packages that use sip use it to generate C code on the fly from sip input files, similar to flex or bison, If a package is reused the same C code on multiple architectures, that's a bad thing [06:42] NCommander: I'm not sure it's a bad thing [06:42] Why would code generation be arch specific instead of generating portable code? [06:42] Well, the other thing about sip which drives me insane is that it doesn't set any default configuration settings [06:43] You can actually ship generated files which you prepared on arch a for people using other arches [06:43] Any package that uses sip must parse sip options and then set then by manual [06:43] Hold on, one problem at a time [06:43] lool, sip generates code for bindings. In some cases, this code is arch specific, i.e, Qt on ARM uses float over double [06:43] right [06:44] So does sip explicitely forbid generating bindings for arm on other arches or vice-versa? [06:44] Nope. It generates invalid code which leads to an FTBFS. This entire bindings issue was caused by the fact that KDE ignored sips options (and the fact that the options file itself didn't properly include the necessary flags) [06:46] other code generation utilities like flex though are the same way; the code it generates isn't promised to be arch independent. I know code generated on ia64 and other architectures thats true (I ran into an RC bug in Debian which this was the case) [06:48] NCommander: I'm trying to find a definitive statement about sip's arch (in)compatibility promise [06:49] * NCommander makes a mental note to replace the wireless router at his mom's [06:49] lool, sorry, the net hiccuped, what was the last thing you said? [06:49] Well, its more the nature of the libraries than sip itself. [06:49] Qt has a different API on different architectures [06:50] Now that API is similar enough that in most cases that you don't see any difference [06:50] NCommander: Well in our case Qt didn't have a different API [06:51] What's our case? O_o? [06:51] qreal :) [06:51] That's still the case [06:51] I didn't change that [06:51] Oh [06:51] * NCommander gets it [06:52] But then even would Qt have different APIs on different arches (which I hope it doesn't), it wouldn't prevent SIP from generating portable code [06:52] It would if you have architecture dependent bits in the bindings code that check for the presence of a feature in sip [06:53] sip wasn't passing the all important configuration flag during compiles to tell it to turn on said feature [06:53] Ok, so you're saying SIP doesn't have the same features on different arches? [06:53] No, I'm not saying that [06:53] What was missing was the C equivelent to an #ifdef [06:54] pyqtconfig has the list of ifdef's programs using sip should use [06:54] (sip itself doesn't read pyqtconfig, it depends on the build system to parse it and give it the right options) [06:55] I'm not convinced; what I see is that there was a runtime check to use a conversion routine; this was made into a runtime + build time check to use it [06:56] sip has no way of knowing the size of qreal at runtime [06:56] unless you want it to do configuration tests on every run [06:56] NCommander: Why can't sip so if (sizeof (qreal) != sizeof (double)) in generated code? [06:56] Or at least ifdef [06:57] The problem with the generated code was it wasn't matching prototypes in Qt [06:57] Essentially [06:57] Besides, #if defined(QT_NO_FPU) || defined(QT_ARCH_ARM) || defined(QT_ARCH_WINDOWSCE) seems like a bad test to me [06:57] lool, that's the test used internally in Qt to switch qreal to float [06:57] That was a copy and paste job :-) [06:57] This test for arches instead of testing for testing for qreal being double or not [06:57] NCommander: I know, and that's why it's wrong [06:58] -"for testing" in the above [06:59] NCommander: I can tell you for sure the #if is wrong; for the rest of the discussion I'm probably lacking a good understanding of what's going on, so I'm afraid I need to defer to someone else or postpone me looking at this right now [07:00] ScottK: Did you follow the sip/kde4bindings/qreal etc. stories; would you be interested in reviewing it? I'm running out of time this morning [07:00] And don't feel like I have enough understanding [07:00] lool: I'm just about to go to bed myself. [07:00] lool, I don't see why using the same ifdef's used by Qt in QtCore.hpp is a bad thing [07:00] ScottK: Would you in general feel confident reviewing this? [07:01] I personally feel I need to learn tons about it first [07:01] lool: No. I've been depending on NCommander's expertise a lot for my condidence in uploading. [07:01] condidence/confidence. [07:01] NCommander: It's a bad thing because it's qt's internal decision; it doesn't regard you how people set qreal to double or float based on the time of the day or the weather outside :) [07:01] lool, I've bugged upstream about fixing the broken configure test so hopefully this will be properly fixed in a soonish upstream release [07:02] lool, fair enough. I did try doing that, but the way sip runs the test causes it to always consider qreal == double [07:02] Probably a header was missing [07:02] hi everyone [07:02] lool: you end up with an x200s? [07:03] jcastro: I wish so, but no nearby stock and didn't want to deliver to the hotel :-/ [07:03] lool: I mean eventually [07:03] jcastro: It's my preferred option so far, definitely! [07:03] :D [07:03] I didn't quite plan how to buy it yet though === RAOF_ is now known as RAOF__ [07:03] I didn't have the money === RAOF__ is now known as RAOF [07:04] jcastro: For your own laptop? Then how did you do it? [07:04] You sold your soul I'm sure [07:04] I didn't have the money for an x200s, had to get the x200 [07:04] Ah [07:04] What's the x200(s)? [07:04] thinkpads [07:05] the s has an lcd backlight [07:05] Exactly [07:05] what's so special about them? [07:05] higher rez screen [07:05] It has the cool things in the x300, but not the crazy things [07:05] Like no SSD and no UWB or what not [07:05] the ssd is an option, they're just kind of a rip off [07:06] Exactly why it was problematic for me to buy a x300: they force the SSD on you [07:06] nod [07:06] fosdem [07:06] Yeah \o/ [07:06] jcastro: You booked yet? [07:06] oops, that's not my google window [07:07] I am looking at it [07:07] jcastro: We should aim for the same hotel [07:07] * Treenaks is thinking of going for one day [07:07] mvo knows the name [07:07] after last year it was so busy that I don't think I would want to make the trip to fosdem [07:07] but this year it's after a canonical sprint so I think on the way back is a good investment [07:07] NCommander: So to close the topic; either I or somebody else need to sit down and see what is currently arch specific and what would become arch specific [07:07] * wgrant was thinking of getting an X200, but it only has a TrackPoint :( [07:08] wgrant: that's a feature. :D [07:08] NCommander: Ideally, nothing is arch specific except qt [07:08] * wgrant is now leaning toward a T400. [07:08] lool, I see [07:08] lool, I'll see if I can find the other python-qt4 maintainer and stick our headers together or something [07:08] NCommander: btw I would like a feedback mail from you, your first UDS, I would like your feedback [07:08] NCommander: Say, sip itself, sip's output, qt, sip's output on arch foo etc. [07:08] jcastro, send me a form [07:09] * NCommander nods [07:09] NCommander: I have no forms, I just want your raw feedback. It can be like "Jorge, I hate your guts ..." [07:10] All and all, I had a good time at UDS, although I disliked having to commute to the Googleplex, and the 128 user limit on the wifi :-) [07:10] jcastro: I'm personally looking at LH4646 for Berlin -> Brussels; the 6th in the afternoon [07:10] yeah well, join the club. :D [07:11] NCommander: I just returned home yesterday, saw a lot in SF last week [07:11] jcastro: I'll toss you a bit of feedback: I didn't like the brainstorm based sponsorship process a bit. [07:12] ScottK: already on the list. [07:13] ScottK: I know exactly what you mean [07:13] OK. [07:13] jcastro: It's not the only reason I didn't ask for sponsorship this time, but it's on the list. [07:44] ScottK, actually I sort of liked it; in past UDS's it always seemed like canonical employees ran all the sessions, and it was really nice to have some that were driven by community members [07:45] if it were opt-in maybe it would have been better received... like have the choice of either being crew, or doing a blueprint [07:45] bryce: I think that's good, but it doesn't need brainstorm for that. In the end UDS is about working on specs, so the brainstorm bit just seemed like more paperwork. [07:46] Right. In then end I wrote specs and didn't go. It seems odd to have writing a spec count for less than brainstorm at a UDS. [07:47] FWIW, I think in Server Team at Prague there was a good mix between Canonical and the Community about who was leading the discussions. [07:47] oh that could be; for prague I did spend most sessions out hacking with timo in the main conf room [07:57] ScottK, what happened to sleep :-) [07:57] Remembered I told my wife I'd do something .... [08:00] ScottK, ah [08:00] ScottK, so don't ask you for sponsoring stuff? [08:03] Not unless it's really simple. [08:06] NCommander: Don't forget you still owe a fetchmail merge. [08:06] :-P [08:08] * popey tries gwibber from jcastros ppa [08:12] NCommander: BTW, I did a run through Intrepid backports today. Then I took a look at Hardy and got overwhelmed. I think it's due for another one of your treatments. [08:13] ScottK - thanks for that, by the way :) [08:13] yeah, I saw that [08:13] ebroder: You're welcome. [13:06] hello [13:06] what is the best way of getting fixes/updates into current ubuntu release [13:07] I want to make Samsung NC10 3G modem usable out-of-the box on 8.10 [13:07] I'm currently working on packaging patches for NetworkManage [13:07] I'm working based on my own nc10 experience and ubuntu community wiki [13:28] damn wine. [13:29] Hobbsee: too much of the good ey? ;-) [13:29] Nafallo: well, it crashed my X when i don't use it as a separate virtual desktop, and then rearranges all my panel icons for me [13:29] ah. WINE. [13:30] yes [13:30] "Feature" [13:31] For the authentic Windows experience [13:31] windows doesn't reorder the icons [13:32] The order of icons in the status area is arbitary, and I've had those re-ordered out from under me [13:32] i was meaning the top panel [13:33] Windows doesn't have one of those :-) === Zic_ is now known as Zic === hyperair1 is now known as hyperair === hyperair1 is now known as hyperair === EagleSn is now known as EagleScreen [17:29] doko_: Around? [17:29] doko_: On armel, /usr/lib/gcc/arm-linux-gnueabi/4.3/libgcc.a has the __sync_* atomic symbols [17:29] But not on my x86 [17:30] It seems to confuse the djvulibre build which uses these === Arby_ is now known as Arby [21:31] jcastro, http://img.ffffound.com/static-data/assets/6/9f6d91778c232d87122e845f5a5e109289bd69d1_m.jpg [21:32] luisbg: hahaha, perfect. [21:33] jcastro, you need to have a tshirt made with this: http://img.ffffound.com/static-data/assets/6/58994c18e6712d014f1cdb4d853b969a7d4805fa_m.jpg [21:33] and that is my dose of awesome of today :) [21:51] When setting up an ubuntu repository, how do I do it without having the source / .changes files? Say I wanted a repo with skype and picasa.. [21:56] ramvi, skype already have thier own repo [21:56] DRebellion: want my own [22:05] ramvi: see dpkg-scanpackages(1) [22:05] apt-ftparchive? [22:17] UTU up again. [22:41] bryce, the last X upgrade is killing me. X starts fine but after a while i get a second mouse cursor, then shortly after that, X crashes. 4 times in the last 24h. [22:44] bryce, it seems like hal is reporting two mouses, one "Macintosh mouse" which i don't have and one "Microsoft Microsoft IntelliMouse? Optical" which i have [22:49] Glad to hear that emgent === Skiessio is now known as SKiessi === SKiessi is now known as Skiessi === cody-somerville_ is now known as cody-somerville [23:48] Hey all, how can I add an installer script to an alternate cd, and force it to use a CLI option rather than GuI option for the system [23:48] ? [23:48] Please [23:54] * WelshDragon test [23:55] * WelshDragon asd [23:55] Hey all, how can I add an installer script to an alternate cd, and force it to use a CLI option rather than GuI option for the system [23:58] easily enough [23:58] the "alternate cd" uses debian-installer, which is trivially coerced into doing things [23:59] generally, you want a "preseed file" which is used by the installer, and contains (amongst other things) a list of packages to install, and a script to run at the end of installation [23:59] A preseed file