=== Cyberworm is now known as Wurm|off === LetoTheII is now known as LetoThe2nd === mckoan|away is now known as mckoan === Wurm|off is now known as Cyberworm [08:35] is preinstalled image still the official way of installing ubuntu on panda? [08:36] hrw: yes https://wiki.ubuntu.com/ARM/OMAP pick server or desktop [08:37] hrw: certainly, or the netboot images. [08:39] thanks === Cyberworm is now known as Wurm|off [09:57] janimo: *pokity poke* [09:58] * ogra_ grumbles about compiz [09:58] alf_, seems they added another quilt change to compiz, there should soon be a new commit so we have to redo everything again [09:59] ogra_: What a great way to start the week ;) [09:59] yeah [09:59] especially isnt i fought over the weekend to even get our friday update reviewed and approved [10:00] * ogra_ wont find the time to do any image testing that way :( [10:00] s/isnt/since/ [10:00] * ogra_ glares at his fingers === brendand_ is now known as brendand [11:23] ogra_: Do you know when they are going to commit the compiz changes? I don't see anything new yet (I am using debcheckout compiz). [11:23] no, waiting for sil2001 [11:23] i guess he will ping me once he is done [11:23] (at least i hope he will) [11:25] ogra_: ok [11:34] Does is exist any one-liner command for installing all unmet build depends for a package? [11:34] apt-get build-dep should do. [11:35] Thats for a source from the apt repo, right? [11:35] of course, it won't work automagically for any src tarball you download from $NIRVANA [11:36] expect IMHO if the tarball is properly debianzied, there might be some trick to do. [11:37] It's a unpacked debian source ready to build. I can always create my little script for it then [11:39] i guess in the debhelper suite or similar i have seen something like that then. but really not sure. [11:40] I remeber debian has a package with a lot of different tools and script for package development [11:40] * sveinse can't remember its name [11:41] ogra_: hi. about the blaze board we discussed last time, did you ship it? [11:41] urgh, thanks for reminding, no i didnt yet [11:41] np [11:41] * ogra_ will make sure to get that dont this week [11:41] really sorry [11:41] if you ship it, i would prefer you ship it to Andy directly (or perhaps someone else at linaro) [11:41] too many compiz updates :P [11:42] yeah, i think thats what we agreed on [11:42] so when you are ready to ship, ask me who to send it to ;-) [12:31] What is multi-arch specifically? [12:34] sveinse: http://wookware.org/talks/ there have a look at those starting with multiarch [12:50] LetoThe2nd: Thanks. Will multiarch be implemented in precise for armel/hf ? [12:52] sveinse: should be, IMHO [13:03] hi guys i'm getting this error any clue? [13:03] pycurl.error: (60, 'server certificate verification failed. CAfile: /etc/ssl/certs/ca-certificates.crt CRLfile: none') [13:07] i got this error while doing : add-apt-repository ppa:tiomap-dev/omap-trunk === suihkulo1ki is now known as suihkulokki [13:44] Hello - I just bumped to Ubuntu Core, a minimal version of Ubuntu [13:45] Is there a possibility that Ubuntu Core would be built for ARMv6 as well as ARMv7? Only the minimal install, minimal amount of packages? [13:46] Walther: from an official point of view, the answer is certainly "no, unless someone pays for it" [13:47] hehe. How many packages are there in ubuntu-core? Approximately [13:47] as in, would it be possible to make it happen as a community project or completely hopeless [13:48] Walther: see https://bugs.launchpad.net/ubuntu/+bug/848154 [13:48] Launchpad bug 848154 in ubuntu "ARM version not supporting V6 RaspPi" [Undecided,Invalid] [13:48] Walther: it might be possible to shape the future.. next UDS happens in two weeks from nw [13:48] now [13:48] Walther: well its open source, the community can do whatever the community wants :) [13:48] it might be possible to lower the optimization grade for armel to support the RaspPi for the next 12.10 release [13:49] since the armel port are now pushed into community hands [13:49] Walther: and i really suggest to look the launchpad thread, it's very interesting. [13:49] LetoThe2nd: Sure, I was just wondering how big effort would it require (certainly much, much less than the full-fletched ubuntu install) [13:50] Walther: and still way more than just using debian. [13:50] Walther: the hardest part are probably providing infrastructure unless canonical nicely keeps the current infrastructure [13:50] Walther: someone have to keep the builders up and running [13:53] i'm using ubuntu-core on pandaboard, even after doing add-apt-repository ppa:tiomap-dev/omap-trunk , and apt-get update , when i do apt-get install ubuntu-omap4-extras , i get unable to locate package [13:54] ...I just got an idea. The first comment on the launchpad page is basically "okay, if we get the resources" [13:54] How about a nice Kickstarter project from Canonical ARM team? [13:55] Given enough pledges, it could be done - and there might be enough people already in the RasPi community alone to provide the necessary funding [13:56] Also, by linking the Kickstarter project to a couple of big tech blogs (e.g. Engadget) would provide additional coverage [13:57] ...and if the goal is not reached, nobody will be charged anything [13:59] Walther: the quickest way to create an armv6 build would be to help the linaro cross compile effort [13:59] Walther: like setup a armv6 optimized toolchain and try cross compile all the packages you need [14:01] Personally, I think a nice Kickstarter project could provide the necessary funds quickly and easily (not to mention, for free) [14:01] Walther: all in all the tasks that ubuntu will do gets decided at the summit next monday in two weeks [14:02] * LetoThe2nd does not believe in kickstarter as the wonder hammer that solves all problems. and especially not when it comes to raspi. [14:02] true. I wonder if anyone here in the channel who happens to work at Canonical could suggest the idea forward [14:02] before the summit [14:02] certainly some people have listened. [14:02] LetoThe2nd: perhaps not a wonder hammer, but it could make the significant difference [14:03] between not making the official build / support and making an official version [14:03] ...oh, and to clarify, I'm talking about (at least) the Core version [14:03] Walther, there is always option f, (fork).. figure you need about 10 pi's backed by harddrives, you can take the src pks of debian/ubuntu and build what you want. ;) [14:03] Walther: well, it all depends. i personally don't see no use in it. but thats my personal opinion. [14:04] I do understand that porting / making support for the complete Ubuntu build would be a big thing [14:04] it would be easier to baseline with debian and use armv4t+ [14:04] LetoThe2nd: Well, if we are talking about *personal* opinions, IMHO Canonical could take advantage of the RasPi and buy a couple thousand and sell as educational bundles [14:05] given there would be support [14:05] then we support all the armv5 devices with 512mb of ram [14:05] that can run a desktop [14:05] Walther, if you want the packages to live in the ubuntu infrastructure you have to port the whole archive [14:05] i don't think the making is *that* big a deal (we've had armv5 and armv6 ubuntus already), but its more like - why care about obsolete, horribly underpowered hardware that is just driven by hype, but not technical reason. [14:05] if you do your own port somewhere else you indeed can do a partial port [14:05] ogra_: perhaps, but even having the Ubuntu Core would be a step forward, right [14:05] no [14:06] LetoThe2nd: you forget the price and educational market [14:06] ubuntu-core is created from the archive ... you woul dneed the full port [14:07] as i said above, you could provide your own archive somewhere with your own build infrastructure (as rcn-ee suggested above) and only build the ubuntu-core packageset indeed [14:07] LetoThe2nd: if you think about it, raspi is 35usd and hs a desktop and office suite for free - compare to the cheapest possible (say 100usd laptop project) + microsoft licenses (Windows, Office, etc) that are used at the moment [14:07] after all though i would suggest debian [14:07] Walther: honestly, to me that is mostly buzzwording. i don't mean to flame here, but giving some kid a thing that isn't able to do much more than just booting into a desktop and that is fully loaded then (but hey, it runs ubuntu!!!) is of little use. better give them arduinos. they have proven worthwhile, and they are even cheaper. [14:07] you wont run a usable desktop on the RPi [14:07] it isnt capable [14:08] ogra_: full ack. i've seen desktops on marvell. forget ever think of really using them. [14:08] right [14:08] ogra_: define capable - there are *many* schools that don't have any computers [14:08] at all [14:08] trying to run a recent browser will make you hit OOM very fast [14:09] they're worthwhile using for some tasks. but certainly not for desktop replacement of any kind (and thats why is implied here), not even by the lowest of standards. [14:09] i bought a raspi so the project survives the startup-phase. the fact, that it is a v6 is surely a showstopper [14:09] current SW simply isnt designed for 256M systems ... you could indeed run something like dillo or w3m ... but i doubt thats what people would call a "desktop" [14:10] LetoThe2nd: replacement? Perhaps no. But replacing not having a computer does make a difference [14:10] i brought a rasppi to twrite tutorials on how to program opengl es [14:10] that i can do without a desktop [14:10] my stance are that debian are good enough [14:10] Walther: well, a few lines above you suggested it as a drop-in replacement for a windows computer including office. and it will just not be that. it will run a cli terminal and emacs fine. but that is probably not what people demand of it. [14:11] LetoThe2nd: No, I wasn't implying a replacement for existing computers with office wtc, I was trying to say that more people would afford that kind of setup [14:11] also dont forget to run a RPi as desktop you will need disks, keyboards, mice and a monitor [14:12] that somewhat wont keep you at the super cheapo level [14:12] Walther: as i said, its all opinions. for me, its buzzwording and wasted effort, until the first raspis are actually brought to real use. [14:12] the RPi is an awesome machine as a PVR or media box [14:12] as well as a NAS [14:13] but definitely not designed for desktop use [14:13] ogra_: but it has 1080p output!!!one!!eleven!! [14:13] And, it is a perfect cheap computer for schools in developing countries that cannot afford more expensive setups [14:13] (it has a great en/decoding engine for media ... but no RAm and no CPU power) [14:14] LetoThe2nd: I never used that as an argument, just sayin' [14:14] Walther, and these schools have monitors, mice and keyboards already ? [14:14] better they would have left out that stupid media chip and invested in ram and cpu power. then it could do things. but of course, it wouldn't be so hypeable. [14:14] 16:07 < Walther> LetoThe2nd: if you think about it, raspi is 35usd and hs a desktop and office suite for free [14:14] (just citing) [14:15] and i said that it will not have desktop and office. not in any usable fashion. [14:15] ogra_: CRT monitors are not only free but paid for if someone takes them, as the process of properly recycling them is expensive [14:15] donating the $25 or whatever your RPi costs nowadays to a project that sends used HW to third world countries is surely a better investment [14:15] ogra_: ack again. [14:16] CRt monitors are heavy ... you need to ship them to that country [14:16] the shipment will be more costly than the CRT is worth [14:16] ogra_: do you happen to know where they are recycled at the moment (at least most of them)? In developing countries... [14:16] (unless you do it in masses in a project like linux4africa for example) [14:16] * LetoThe2nd is out again. point has been made clear. [14:17] Anyway, I think we are getting to a sidetrack of opinions vs opinions [14:17] well, to come back to the topic, ubuntu doesnt roll the archive after an image seed but rolls images from a complete archive [14:17] sticking to the original thing - what would it require and what would be the easiest way of getting some extent of Ubuntu [14:17] so what you asked for wont be possible in the current infrastructure [14:18] either do your own infrastructure and roll your own port or convince canonical to invest into an armv4/v5/v6 port [14:19] the latter would indeed mean a full archive port [14:19] and that will take you 6 months with a team of fulltime people [14:20] (plus HW ressources, the current ubuntu arm build machines are fully saturated with what we have already) [14:21] Oh, and another thing - iirc, Canonical has the Ubuntu TV project [14:21] it does [14:21] Raspi would be a good possibility for that - XBMC has been proven to work smoothly on RasPi [14:21] unity-2d wont run in 256M [14:22] Walther, playing video and using the desktop are two different things, when you have a hardware video decoder.. [14:23] as i said before, the RPi will make an awesome PVR/DVR or media playback machine [14:24] but you wont run a desktop on it in any usable way [14:24] invest a few bucks more and buy a cubox or so if you want a desktop ;) [14:24] and of course, is the video stack open? ;) if not, is anyone able to support it.. [14:24] Is Ubuntu TV intended to be a *full* desktop? [14:25] ubuntu tv is an extension to unity-2d [14:25] I though it was going to be just that, a PCR/DVR with Ubuntu features [14:25] PVR* [14:25] its just an additional lens that handles media and epg data [14:26] so an official Ubuntu TV device would indeed run a full Ubuntu desktop (with unity-2d, that is)... [14:27] an official ubuntu-tv device would be a TV [14:27] well, that's quite a lot of requirements imho for what is essentially a PVR/DVR [14:27] not less than google-tv [14:28] and not less then your avg, mythtv system. ;) [14:28] yeah [14:31] Anyway, so what was the linoros thing you mentioned briefly in the beginning? [14:31] some kind of cross-compiled build someone is doing [14:31] see #linaro [14:32] thanks, this seems to be what I'm after [14:35] Walther, a lot of this is just talk... There's enough of you guys that want, please pull together, and follow the gensi example.. Before the notion of ubuntu/debian armhf, they did their own 'armhf' port of debian, and then showed what was possible with it.. They then worked with the community, then after some time, it's now in debian/ubuntu as an port.. [14:35] yep, I'm not going to give up [14:36] and I'm not interested in flamewars or similar, opinions are opinions - even good, supported opinions are just opinions that can be opposed with other good, supported opinions [14:37] and yes, I believe there are enough people in the Raspi community that at least some kind of a build will be made at some point [14:37] well samsung smart tvs run an openembed of sorts, mine has a fairly powerful CPU [14:38] Walther, figure you'd need 5-10 pi's.. just port the intial 'debootsrap' requirements, and get it on: http://www.debian-ports.org/ [14:39] rcn-ee: I think cross-compiling would make more sense - iirc it took about 3h to compile the quake III port :P [14:40] Walther, you cant easily cross build something with a big dependency chain [14:40] Walther, except that's against the debian/ubuntu where everything is built native.. ;) [14:40] building natively makes more sense in that context even if it takes a bit more time [14:41] if you only build a single app, cross is surely a good thing [14:41] if you want cross-compiled distro, look at angstrom... [14:41] but try to cross build a desktop :) [14:41] ogra_: possible, but difficult indeed. [14:41] yeah, for cross use a distro thats designed for cross from the ground up [14:41] like openembedded/angstrom etc [14:42] I build my kernels cross compiled because it saves a few hours, but I would hate to build more than that [14:42] .oO( which remonds me, i need to add the embedded spec to the UDS tracker ) [14:42] *reminds [14:42] lilstevie, hours ? [14:42] wow [14:43] * ogra_ builds the ac100 kernel package in about 1.5h natively [14:43] admittedly on a fast external USB disk, not on the emmc [14:44] ogra_: the asus kernel on the tf101 and tf201 have some horrible bottleneck somewhere [14:45] trimslice is slghtly faster [14:45] get an ac100 ;) [14:45] it flies :) [14:45] (hehe, compile times are silly... I remember compiling a linux 3 kernel in 45s) [14:45] (if you throw it at least :P ) [14:46] heh well builds take 20minutes on my core duo [14:46] my panda did one in 9mins last night... oh the wonders of ccache. ;) [14:47] thats cheating !!! [14:47] and the incremental build with a single non-important line change helped. ;) [14:47] and I haven't tried a compile since the migration to 3.1 on the tf201 [14:48] the new kernel is faster [14:58] Yeah, well, I was abusing a server I'm on to achieve the sub-minute compile time [14:58] 24 cores on a server with 144GB ram [15:03] you don't need anything nearly that expensive to build kernels in under a minute. [15:05] Nope.. [15:05] Although did you notice that the current kernel broke distcc? [15:05] I can't get away with a multi-machine compile at the moment [15:08] Quantal Quetzal ! [15:09] ogra_: PP... ;) [15:24] GrueMaster: Regarding your 'Community support request' to help testing beagle boards [15:25] is this the daily url http://cdimage.ubuntu.com/ubuntu-server/daily-preinstalled/current/ ? [15:25] cehh, yep [15:25] cehh: Yes. [15:25] and http://iso.qa.ubuntu.com/qatracker/milestones/214/builds for tracking your results [15:26] OK, thanks. [15:26] GrueMaster: when you have time, please reply to my question on BeagleBoard [15:27] I basically need to know if it is possible to add test tools (iperf, lmbench, etc. ) to the filesystem image [15:28] cehh: Yes, after booting. You should be able to run "apt-get install" to install most of the ~15,000 packages in the Ubuntu pool. [15:28] Hi what happened to http://people.canonical.com/~tobin/natty/beagleXM-natty.tgz [15:28] Whether or not they actually work is another question. [15:29] beagleboarduser: It went away when I did. Sorry. Switch to Oneiric (11.10) or Precise (12.04 - now in final release testing). [15:30] is the 12.04 very buggy? [15:30] the opposite :) [15:30] very good [15:30] will try 12.04 [15:30] beagleboarduser: I hope not. Release is Thursday. [15:30] its more like a street car. [15:30] will try it anyways [15:36] cehh: Could you send me a link to the beagleboard topic? Otherwise I will have to wait for the daily summary. [15:36] GrueMaster: https://groups.google.com/forum/#!topic/beagleboard/3NFPUue_oG4 [15:36] thx [15:37] btw, is the kernel update documented at https://wiki.ubuntu.com/ARM/Server/Install?action=show&redirect=ARM%2FOMAPHeadlessInstall still required w/ latest server image? [15:46] cehh: That was a one-image only update due to new hardware that came out at release time. Support was added to the next Ubuntu release as native. [15:46] (and the links are probably broken now). [15:49] got it. One less step to do :) [15:51] I will need to create an apt package for ltp-ddt. We currently use OE to build it. The other test tools that we use seem to be already available with apt [15:54] ...just wondering, is Ubuntu ARM compatible with Marvell 88AP510 SoC [15:55] Walther: see the topic [15:58] ...could you clarify which part of it answers to my question? [15:59] Walther: is 88AP510 armv7 or not? [16:01] Marvell Armada 510 (88AP510) SoC with an ARM v6/v7-compliant superscalar processor core [16:01] so not really sure [16:01] so in v7 mode it will :) [16:02] 'kay, thanks [16:02] cubox comes with armada 500, thats definitely working [16:02] wasn't that marvell thing some homebrew v6/v7 schimera? [16:02] yep [16:02] marvell does such things [16:03] *notes* if i ever want to brew schimeras, apply at marvell. [16:03] hehe === mckoan is now known as mckoan|away [16:05] * ogra_ thinks schimeras are a good tradeoff if you also get GigE, PCIE and proper SATA by default on their boards :) [16:06] did i say anothing bad? [16:06] nah [16:27] ogra_: ping... [16:28] djszapi, yes ? [16:28] What development package installs linux/rtc.h on my ubuntu ? It is installed by linux-api-headers on my Archlinux box. [16:28] use dpkg -S [16:28] (with the full path) [16:29] which package to install: [16:29] 1) linux-headers-3.0.0-1208-omap4 [16:29] 2) linux-headers-3.1.1-26-linaro-lt-omap [16:29] the one matching your used kernel [16:29] djszapi: linux-libc-dev [16:29] ogra_: the linaro one, thanks. [16:29] ubuntu@panda3:~$ dpkg -S /usr/include/linux/rtc.h [16:29] linux-libc-dev: /usr/include/linux/rtc.h [16:30] weird, the linary header is already installed, but I still have zero linux/rtc.h :/ [16:30] well, what GrueMaster said [16:46] thanks ogra_ and GrueMaster [16:55] hi, i've been trying to cross compile a simple program using arm-linux-gnueabi-* tools on ubuntu 11.10 . but despite of -march=armv6 switch its is always producing armv7 executable. [16:57] how can i make it produce armv6 code ? [16:59] ssilly: The ubuntu build tools are hardwired to only produce armv7. You would need to rebootstrap gcc and the other low level tools and libraries to do this. You might want to consider Debian for this type of work. [17:00] well, theoretically you should be able to override the default target arch [17:01] but you probably need to set more than just -march [17:01] i guess infinity might know more as a toolchain expert [17:03] thanks, so i should either compile the toolchain , or go with debian and hope that it works .. [17:03] debian armhf gcc uses the same defaults [17:07] oh. so basically i cannot skip the gcc compilation :( [17:07] armel debian uses v4t [17:10] thanks. i'll give debian a try [17:11] ogra_: What is the easiest way to upgrade ac100? Can I dd the bootimg to a partition and let it take over on next reboot? [17:11] thats what i do usually [17:12] though i would suggest to also dd the sosboot image to the recovery partiton [17:12] Ok. What is the partition for dd? [17:12] then, even if you screw up, you can enter an initrd in the recovery partition [17:12] And where is the sosimage? [17:12] check with abootimg -i [17:13] GrueMaster, http://ac100.grandou.net/sosboot [17:24] Ok, installing. cool. === Wurm|off is now known as Cyberworm [17:53] ogra_: Did the nVidia driver for AC100 armhf make it into the pool? [17:54] nope [17:54] it didnt make it our of nvidia yet :) [17:54] *out [17:54] ah. [17:55] Hmm. Hash sum mismatch on ports.ubuntu.com/ubuntu-ports precise main sources. [18:11] It would be nice to have hdmi out this weekend. I am heading to LinuxFestNW (http://linuxfestnw.org). [18:11] well, then you would have to use armel [18:12] But I want it all! :P [18:13] heh, call nvidia then :) === arun__ is now known as arun_ [19:32] GrueMaster requested testers for the beagle board xm. Should i be using the daily images or something more specific? [19:32] daily is fine [19:32] and log results at http://iso.qa.ubuntu.com/qatracker/milestones/214/builds [19:32] looking at the qa tracker it looks as if no one has really updated the xm [19:33] i will update my findings tonight [19:33] or omap at all [19:34] awesome, thanks ! [19:37] ogra_: do i need a launchpad account to post? [19:37] yep, i think so === Matt_O1 is now known as Matt_O === NekoXP is now known as Neko [21:30] infinity, peekity peek [23:25] booting the daily image on a beagleboard xm, there is no partition option available [23:26] is this expected or is there a way to load gparted?