**** BEGIN LOGGING AT Fri Mar 09 02:59:57 2007 Mar 09 05:50:48 I could use some help debugging something gone wrong down deep in MokoMakefile. Mar 09 05:51:16 I'm hitting a failed/missing configure substitution here: Mar 09 05:51:19 | make[3]: Entering directory `/hda3/moko/build/tmp/work/i686-linux/gettext-native-0.14.1-r4/gettext-0.14.1/gettext-runtime/po' Mar 09 05:51:19 | /bin/sh @MKINSTALLDIRS@ /hda3/moko/build/tmp/staging/i686-linux/share Mar 09 05:51:28 ... Mar 09 05:51:30 | FATAL: oe_runmake failed Mar 09 06:00:26 Me hacks Makefiles, replacing "MKINSTALLDIRS = @MKINSTALLDIRS@" with "MKINSTALLDIRS = $(top_srcdir)/../config/mkinstalldirs", but wishes he could contribute a proper fix somewhere. Mar 09 06:00:35 /me even Mar 09 06:17:53 I seem to be getting something similar to the 8250 vs pxa namespace conflict on linux-rp-2.6.20.. Mar 09 06:24:49 cworth: so don't use the mokomakefile Mar 09 06:25:17 koen: Oh? Is it being unhelpful to me? Mar 09 06:27:14 Is this a known problem with the pxa270 based stuff? Mar 09 06:28:54 03mwester 07org.oe.dev * rc93190f5... 10/ (1 packages/meta/slugos-packages.bb): slugos-packages: mediatomb removed as it will not build either. Mar 09 06:28:58 03mwester 07org.oe.dev * rd446e3e6... 10/ (1 packages/meta/slugos-packages.bb): slugos-packages.bb: demote mpd, breaks the autobuild too. Mar 09 06:31:08 cworth: the makefile just sets up your env and is meant for people too lazy to read how to setup OM/OE Mar 09 06:31:46 koen: Hey, don't knock my laziness! Mar 09 06:32:11 cworth: and your problem is with the gettext-native package, not with the mokomakefile Mar 09 06:32:20 koen: But yeah, I should probably look deeper here. Mar 09 06:32:26 I suspect something on your host is throwing in a spanner somewhere Mar 09 06:32:35 koen: Yes, I presumed as much, (which is why I came here to ask about it). Mar 09 06:32:38 (-native packages get built with host tools) Mar 09 06:32:50 koen: Ah, I see. Mar 09 06:33:18 koen: Are you a good person with whom to debate switching from monotone to something like, say, git? Mar 09 06:33:35 heh Mar 09 06:33:57 cworth: we've had 2 switching 'points' already and no-one wanted to swtich from monotone Mar 09 06:34:19 lots of shouting and saying "OMG GIT!!!" but in the end all from complainers-not-doets Mar 09 06:34:22 doers* Mar 09 06:36:51 cworth: and we have good monotone support, since one of our developers is also a monotone developers Mar 09 06:37:01 and some other monotone developers hang out here Mar 09 06:37:05 morning all Mar 09 06:37:07 koen: Ah, I see. Mar 09 06:37:10 hey Ifaistos Mar 09 06:37:30 koen: Well, I've already run into a fair amount of monotone pain with just trying to get started with oe. Mar 09 06:37:41 no doubt Mar 09 06:37:57 but I don't expect git to "just work" for everyone Mar 09 06:37:58 koen: And I imagine it would be less work for me to import this stuff into git and work with it there, than for me to try to learn how to work with monotone. Mar 09 06:38:18 koen: So, if it would be interesting for people to look at what that would look like, I could give that a try. Mar 09 06:38:31 * koen looks at the numerous bug reports of "can't fetch git packages" due to debian git vs git-core Mar 09 06:39:25 * cworth looks at the horrifying mess that is monotone.ca/UPGRADE Mar 09 06:40:03 cworth: A lot if the devs keep a "local" copy of OE with their favorite tool, but you have to get somewhat familiar with monotone one way or the other Mar 09 06:40:07 Someone mentioned that oe selected montone over git since git was "alpha" software at the time, but those UPGRADE instructions suggest to me that monotone still is. Mar 09 06:41:14 cworth: i am using svn for our local copy but have to move things in/out if you want to keep in sync (at least in the parts that matter) Mar 09 06:41:31 cworth: classes, build tools, some core packages etc Mar 09 06:42:05 Ifaistos: Sure. Much easier to just make the clean break and switch everything to git of course. ;-) Mar 09 06:42:09 git couldn't even manage to import the flat file tree (no history!) when we first tried it Mar 09 06:42:43 (cogito couldn't either) Mar 09 06:42:51 koen: Was that git dying or some import tool? (That's a huge difference...) Mar 09 06:43:08 plain git and later git + cogito Mar 09 06:43:20 that was during the bitkeeper days Mar 09 06:43:20 Ha. I have revisions in one of my mtn databases from 3.5 years ago. :) Mar 09 06:44:19 cworth: the most practical reason against git is that a number of core developers absolutely hate git and would probably stop contributing after switching Mar 09 06:45:07 cworth: well i was also a bit reluctant with mtn but on the other hand we learn to use a gazilion of tools to do our job. A gazilion+1 would not make a difference Mar 09 06:45:45 and you can still kill someone's local repo when doing 'git pack' on the server at the wrong time Mar 09 06:46:01 data loss rules! Mar 09 06:46:05 koen: No, that's not true. Mar 09 06:46:36 koen: At least not with current git. The current repack stuff is totally cron-job safe. Mar 09 06:47:24 koen: have you been able to figure out why gcc brakes on efika/ppc ? Mar 09 06:47:24 Ifaistos: I'm willing to learn new things. It was just that my first introduction to monotone was giving up after waiting over 1 hour to get halfway through a download of something like 10 MB or so. Mar 09 06:48:01 Good morning chat :) Mar 09 06:48:15 koen: it's *very* strange Mar 09 06:48:26 Ifaistos: not yet Mar 09 06:48:35 and I'm leaving for the airport in a few minutes Mar 09 06:48:41 koen: And as for data loss, my other first introduction to montone was what the openembedded pages say about it: "monotone db migrate command fails on some updates". That's scary stuff. Mar 09 06:49:14 cworth: I think there was a time a little while back when there were a bunch of merges in a short time,which really slowed down the pull. Mar 09 06:49:22 koen: hehe this means pleanty of time to debug it on flight :) Mar 09 06:49:42 * hvontres|home imagines sitting next to koen and his toys on a plane... Mar 09 06:50:03 hvontres|home: When I asked about it, I was given several stories about having to change to use the magic version of montone to make it all go fast. I didn't succeed at following all the details. Mar 09 06:50:31 hvontres|home: This was only a couple of days ago. Mar 09 06:51:16 with the security checks these days you have to make sure that all your gadgets at least power up ;) Mar 09 06:51:28 cworth: how old was the db snapshot you were using? Mar 09 06:51:38 or you might loose your flight Mar 09 06:52:08 Anyway, I'll see if I can find a reasonable montone->git conversion tool andtry to do some comparisons on total storage and speed of checkouts, etc. Mar 09 06:52:18 Ifaistos: how would you proove to the security guy that your naked dev board actually powered up????? Mar 09 06:52:24 hvontres|home: I have no idea. Mar 09 06:54:20 cworth: I haven't had a whole lot of mtn issues on my poor little old machine, but I only use it for ro access. But 0.33 does seem to be pretty good these days. Mar 09 06:54:39 cworth: and I am running this on an old dual PII-333 :) Mar 09 06:55:24 hvontres|home: Well, git will be strictly faster than monotone I'm quite sure. Mar 09 06:55:27 hvontres|home: yeah i know what you mean.... had to go to the states in Nov with a bunch of boards and had to go through the routine 3 times Mar 09 06:55:41 cworth: 'faster' != better Mar 09 06:56:01 koen: OK, faster, smaller storage, more robust, ... Mar 09 06:56:18 still not reason to switch Mar 09 06:56:40 flat files on a webserver is even faster Mar 09 06:56:53 cworth: well IMHO people made their decision to use this tool and i think it will take something major to change Mar 09 06:56:58 koen: Wait, what? Mar 09 06:57:11 koen: I thought you just said "'faster' != better" ? Mar 09 06:57:28 Ifaistos: I guess that would give you an incentive to get a bootable image done BEFORE you go :) Mar 09 06:57:43 koen: Oh, or were you saying that one could design a system that's worse that's even faster. Mar 09 06:58:14 cworth: so just go the flow as far as commits go and keep a local copy for daily use with your favorite tool Mar 09 06:58:20 cworth: I think that's what he was trying to say. Mar 09 06:59:58 hvontres|home: well i had spend a few months to get them working and i was not under any circumstances check them in with the rest of baggage Mar 09 07:00:43 hvontres|home: i could manage without clothes for a few days... but without them it will be like not having been there Mar 09 07:01:06 * Ifaistos goes to get some coffee Mar 09 07:01:08 Ifaistos: heh..been there, done that. Mar 09 07:01:57 So, how much branching does oe have inside its monotone stuff? Mar 09 07:04:26 * Ifaistos says there is nothing like the smell of fresh coffee in the morning...except napalm Mar 09 07:05:04 * hvontres|home thinks napalm might burn a bit going down Mar 09 07:05:40 hvontres|home: You're only supposed to smell it! Mar 09 07:06:45 Weird. My kernel patch actually worked. :) Mar 09 07:06:48 hmmm....top - 23:06:14 up 1:50, 0 users, load average: 8.46, 7.78, 6.52 Mar 09 07:07:12 hvontres|home: you have an 8 way system, I hope. :) Mar 09 07:07:15 methinks running two bitbake sessions on my old hw isn't the best idea Mar 09 07:07:29 Necronom: nope...dual PII-333 :) Mar 09 07:10:04 hvontres|home: i had Celeron P4 2.5Ghz overclocked to 3.5 and when i got to builds running it start sending kernel messages for high temperature :) Mar 09 07:10:25 to->two Mar 09 07:10:49 bitbake and OE another way to stress test your system Mar 09 07:12:06 * hvontres|home just checked to make sure I wasn't getting stuff like that... Mar 09 07:12:34 koen : Probably in the weekend ,I will make a document on how to get Angstrom going on Efika, would it be possible to put it on the Angstrom site ? Mar 09 07:14:20 koen [n=koen@dominion.kabel.utwente.nl] has quit [] Mar 09 07:16:16 cworth: we do not have much branching inside the OE trees, AFAIK Mar 09 07:16:48 likewise: Yeah, I found a list of a handful of branches only. Mar 09 07:17:24 cworth: and good to see you here. I come by #cairo to little. Mar 09 07:17:27 too Mar 09 07:17:33 So that should be good, (as I think that the tailor conversion tool only handles a single branch, but that should still be significant enough to make for a reasonable comparison). Mar 09 07:17:57 likewise: And likewise, it's nice to see you here too, ... likewise. Mar 09 07:20:36 Re Mar 09 07:26:53 Ifaistos: ping. What do you mean exactly by running Angstrom on the EFIKA? The graphical part of it? Mar 09 07:29:38 cworth: I think the only really active branch is .dev right now... Mar 09 07:29:42 likewise2: Well how to get Angstrom running on your Efika :) Get the image, put it in a usb stick etc Mar 09 07:30:58 Ifaistos: ok, but there is nothing special to it, is there? Do you have X support going etc? Mar 09 07:32:08 likewise2: Yes, i am running Angstrom on it for sometime now :) I was asked to documented by the Efika guys Mar 09 07:34:21 Ifaistos: OK, yes, would be nice :-) Mar 09 07:34:35 So, is this really the simplest command line for updating the oe stuff with monotone? Mar 09 07:34:40 mtn --db=../OE.mtn pull monotone.openembedded.org org.openembedded.dev; mtn update Mar 09 07:34:56 cworth : nop Mar 09 07:35:28 cworth :once you have checkout a local copy an mtn pull, mtn update is enough Mar 09 07:35:42 Ifaistos: Tell me the shortcut then, (and let's fix the "Updating OE" part of the GettingStarted wiki page). Mar 09 07:36:03 cworth : The line you mention is if you have multiple db's on your system Mar 09 07:36:33 Ifaistos: So what is the pull spending 45 seconds doing? (And why on earth does it take 45 seconds for it to realize it has nothing to do...) Mar 09 07:36:57 | Installing ipkg (0.99.163-r2) to root... Mar 09 07:36:57 | Downloading file:/stuff/tmp/deploy/glibc/ipk/mipsel/ipkg_0.99.163-r2_mipsel.ipk Mar 09 07:36:57 | Configuring ipkg Mar 09 07:36:57 | (offline root mode: not running ipkg.postinst) Mar 09 07:36:57 | An error ocurred, return value: 4. Mar 09 07:36:58 | Collected errors: Mar 09 07:37:00 | ERROR: Cannot satisfy the following dependencies for task-base: Mar 09 07:37:02 | update-rc.d update-rc.d makedevs update-rc.d sysvinit-inittab sysvinit-pidof tinylogin update-rc.d setserial update-rc.d sysfsutils module-init-tools update-rc.d update-modules module-init-tools sysvinit-pidof ipkg-collateral util-linux-mount util-linux-umount tslib-tests tslib-calibrate gdb strace procps Mar 09 07:37:06 | Cannot find package ipkg-collateral. Mar 09 07:37:08 | Check the spelling or perhaps run 'ipkg update' Mar 09 07:37:10 NOTE: Task failed: /stuff/tmp/work/db1200-angstrom-linux/bootstrap-image-1.0-r0/temp/log.do_rootfs.17130 Mar 09 07:37:16 Ifaistos: So if I do "mtn update" in my checkout it will or will not update the db I made that checkout from? Mar 09 07:37:28 how can i "ipkg update"? Mar 09 07:37:32 cworth: it will not Mar 09 07:38:06 likewise: So there's a different database that it's updating? Mar 09 07:38:18 cworth : You do a pull to get data in your local mtn db, then you do an update to update your checkout copy Mar 09 07:38:31 cworth: the .db is your local repository, you can add or commit to it. Mar 09 07:38:32 Ifaistos: Right, that's what I understood. Mar 09 07:38:58 So, if I need to get new changes that have been committed remotely, then I need to do the pull and update I showed above, right? Mar 09 07:39:09 cworth: correct Mar 09 07:39:54 cworth : but you don't need to add all the stuff just an "mtn pull" is enough Mar 09 07:40:09 So, my questions are: 1. Is there really not a simpler command for the pull? Does it not have any way to have me store a simple name for a canonical upstream source to pull from? Mar 09 07:40:28 2. You all can really stand waiting up to a minute for a do-nothing pull? Mar 09 07:40:29 cworth: *if* you are in the checkout directory, then yes, you can omit all the details. Mar 09 07:41:11 likewise: Ah, so a "pull" from a checkout directory _will_ update the database and _does_ know where to do the pull from. Is that it? Mar 09 07:41:51 likewise: Oh, but is it only pulling a single branch then? Mar 09 07:42:17 cworth: yes to the 1st q, because a checkout remembers the repository database you are working with and monotone assumes you meant that one. Mar 09 07:42:29 likewise: That is, in the case covered on GettingStarted where I've got two different branch checkouts, would I have to do a pull in each to get all the updates? Mar 09 07:42:56 cworth: monotone itself has a very good manual (MUCH better than GITs :-) ) Mar 09 07:43:43 likewise: I don't quite follow that entirely. I created the database without monotone being aware of the source at all, (I just obtained it with wget). Then I pulled into it once from oe.org. Then, I created a checkout from the local database. Mar 09 07:44:12 So, why does the checkout remember that a pull should come from oe.org, but the database itself does not? I'm still confused. Mar 09 07:46:12 Anyway, the shortcut "mtn pull" is really a minor issue. The major issue is what on earth is it doing that it takes it so long to realize that there's nothing to pull?! Mar 09 07:46:32 cworth; you might be interested in "mtn ls vars", and default-include-pattern. Mar 09 07:47:01 cworth: I see your point, let me look at the docs (I am a mere developer with commit access using monotone, I did not specificly select it). Mar 09 07:47:21 Necronom: Ah, very nice, thanks. Mar 09 07:47:47 Necronom: So that shows that the db does know oe.org as default-server, and sure enough, this works fine: Mar 09 07:47:54 mtn --db=OE.mtn pull Mar 09 07:48:10 So, if nothing else, the GettingStarted instructions could be simplified a bit. Mar 09 07:48:44 cworth: if you want to pull multiple brances you can do something like mtn pull --db=OE.mtn monotore.openembedded.org org.openembedded.{branch1,branch2,...,branchN} Mar 09 07:49:35 Does somebody have a problem with Xfbdev ? When i start it giving segmentaion fault. That is with X11R7.0 Mar 09 07:50:14 Sorry with X11R71 Mar 09 07:50:17 Anyway, I should sleep now. I'll try to convert this OE.mtn to git tomorrow and see what comparisons I can make. Mar 09 07:50:21 Thanks, everybody. Mar 09 07:50:27 cworth: Yes, the OE learning curve is a bit steeper than that I have seen with the other cross building projects (scratchbox, crosstool, buildroot, T2, LFS, ...) but once at the top you'll love it Mar 09 07:51:48 likewise: Well, this is all giving me something to do while I wait for glibc to build of course. We'll see how far it has all progressed by morning... Mar 09 07:52:23 Though it looks like I might not have sufficient disk space on the partition I put it in. How much does it usually end up taking? Mar 09 07:52:54 For my embedded targets, up to 3 GB if I leave all the source directories and build dirs floating around Mar 09 07:53:46 cworth: there is a bitbake (build recipe) class called rmwork that removes sources/build dirs after each package was built, but it is not inherited by default. Mar 09 07:54:15 likewise: Hmm... OK. I put it in a 5GB partition, but it's 48% full and building glibc which gets me a bit nervous... Mar 09 07:55:08 Ah well, I guess I'll get to see if it has failed in the morning. Good night. Mar 09 07:59:43 good night Mar 09 08:00:52 i am in endianess hell Mar 09 08:01:01 Ifaistos: big or little? :-) Mar 09 08:01:04 ~lart endianness Mar 09 08:01:04 * ibot teaches endianness the basics, including how to RTM Mar 09 08:01:13 both Mar 09 08:01:15 ~hail big Mar 09 08:01:17 * ibot bows down to big and chants, "I'M NOT WORTHY!!" Mar 09 08:03:13 have you seen an *fast* code to do conversions ? Mar 09 08:15:47 * hvontres|home wonders why we wound up with two versions of endianness... Mar 09 08:18:34 actually there are 3 Mar 09 08:19:09 there is also middle endian Mar 09 08:20:21 The good about standards is that you have so many to choose from and make your system incompatible with everything :) Mar 09 08:22:13 hehe...and you can always make up a new one in a pinch..:) Mar 09 08:22:46 Ifaistos: at least we don't have to worry about inch bits vs metric bits :) Mar 09 08:24:18 well, the day EU will decide on a common standard for wall plugs, then we will be really a Union :) Mar 09 08:25:27 Ifaistos: heh..there is always the "Europlug"... the one plug guaranteed to fall out of every socket in europe :) Mar 09 08:26:11 * hvontres|home wishes inches would finally just up and DIE like they should have 150 years ago.... Mar 09 08:30:05 how many thumbs go in a foot? Mar 09 08:30:29 likewise: no clue :) Mar 09 08:30:35 :-) Mar 09 08:31:01 I think they decided on twelve :-) Mar 09 08:31:58 likewise: But my favoritie anachronistic unit of all times: The Ton of Refrigeration= the amount of energy needed to melt one ton of ice in one day... Mar 09 08:32:19 and it's still used to size AC units... Mar 09 08:34:24 well, time to go to bed :) Mar 09 09:09:10 is there any althernative to xserver-kdrive ? Mar 09 09:10:07 I can'n run it, on start Xfbdev always gives me seg fault Mar 09 09:10:48 is there any other reports for this problem ? Mar 09 09:11:09 <_law_> lkm`, i think there is also xserver-org in OE Mar 09 09:11:42 i need something for armv4 Mar 09 09:13:37 it seems that problem is only in Xfbdev, because i can run all other xbase application, if i export DISPLAY to other xserver Mar 09 09:20:13 cworth: ping Mar 09 09:20:18 morning Mar 09 09:22:00 cworth: "mtn pull" has "--set-default" option which will store information about which monotone server to pull from and which branches. Mar 09 09:25:08 cworth: and 'mtn pull' have to get info which certs are on server from which it has to pull because this is distributed SCM and you can pull from misc servers Mar 09 09:27:11 cworth: and with monotone I *know* how much data it has to fetch. with git/cogito I can only hope that it will end soon Mar 09 09:27:48 cworth: and empty mtn pull here isreal 0m13.541s Mar 09 09:27:48 user 0m6.967s Mar 09 09:27:51 sys 0m0.192s Mar 09 09:27:52 ops Mar 09 09:46:23 good morning all Mar 09 10:13:47 good morning Mar 09 10:16:48 hi florian Mar 09 10:36:55 here is my "bogofeed" implementation: http://pastebin.ca/387413 Mar 09 10:36:57 any comments? Mar 09 10:38:18 rootfs_ipk_do_indexes before copy Mar 09 10:38:27 so you will get rid of old packages first Mar 09 10:38:35 ok, done Mar 09 10:38:54 I also removed the keep packages if around the last make-index, cause it will always be created from scratch Mar 09 10:39:03 and I would do cp ${DEPLOY_DIR_IPK}/$d/*.ipk ${DEPLOY_DIR_IPK}/bogofeed/ instead of find Mar 09 10:39:37 same effect, less seeking Mar 09 10:39:59 ah, but you get arguments too long on big feeds. Mar 09 10:40:15 updated: http://pastebin.ca/387415 Mar 09 10:40:52 yes.. Mar 09 10:41:07 do I need to exclude morgue dirs somehow? Mar 09 10:41:26 remove them after indexes Mar 09 10:41:45 as find will go into them.. Mar 09 10:41:48 and I guess this all needs to be wrapped in a ${DEPLOY_BUILD_BOGOFEED} conditional Mar 09 10:42:36 rwhitby: or do: cp `ls ${DEPLOY_DIR_IPK} -1|grep ipk` ${DEPLOY_DIR_IPK}/bogofeed/ maybe? Mar 09 10:45:18 no good if it's empty. Mar 09 10:46:00 updated: http://pastebin.ca/387418 Mar 09 10:49:15 rwhitby: you can also split locales into subdirs Mar 09 10:49:39 ooh - hang on, you're adding requirements now. Mar 09 10:50:12 no feature creep please. this is simply getting back to status quo. Mar 09 10:51:44 no other comments? I'll send it out for RFC then. Mar 09 10:52:41 no more Mar 09 10:55:01 sending RFC now Mar 09 10:56:14 hrw: In which .bbclass should I set DEPLOY_BUILD_BOGOFEED to "0" by default? Mar 09 10:56:36 bitbake.conf? Mar 09 10:57:46 just after DEPLOY_DIR_TOOLS ? Mar 09 11:02:17 yes Mar 09 11:02:39 So OE has a strict requirement for Python 2.4 at least now. Mar 09 11:02:39 (due to the use of rsplit in package QA Mar 09 11:03:51 was that intentional? Mar 09 11:04:41 no idea Mar 09 11:04:55 ah, yes. it's clearly documented on http://www.openembedded.org/wiki/RequiredSoftware Mar 09 11:05:15 Unfortunately, that rules out RHEL4, but that's my problem now, cause it's clearly documented. Mar 09 11:10:19 anybody did a bb meta-sdk recently? it fails with "ERROR: No providers of runtime build target libqpe1" on bb 1.8 Mar 09 11:10:39 chris144: remove that lib from it Mar 09 11:11:15 chris144: its pda styled sdk ;( Mar 09 11:11:31 chris144: build task-sdk-base Mar 09 11:11:39 hrw: ok; is it worth submitting a patch in bugzilla? or is it still needed for pda stuff? Mar 09 11:12:15 chris144: I think that it would be good to remind that problem on ML. we need to have SDK targets I think Mar 09 11:13:05 hrw: absolutely aggreed, I will test meta-toolchain too before sending to the list... Mar 09 11:18:07 Why is "FEED_URIS_append_familiar" in bootstrap-image.bb ? Mar 09 11:18:45 03florian 07org.oe.dev * re9800ff1... 10/ (3 files in 3 dirs): linux-ezx: Add additional fix to improve the touchscreen behaviour. Mar 09 11:18:50 03florian 07org.oe.dev * ra493ab6d... 10/ (1 conf/machine/fic-gta01.conf): fic-gta01: Pad to the end of flash to work around filesystem issues due to lacking flash erase. Mar 09 11:25:12 can someone look at packages/angstrom/angstrom-mokogateway-image.bb, and tell me why it shouldn't be a tasks/task-mokogateway.bb combined with image-bootstrap.bb ? Mar 09 11:29:36 http://www.dilbert.com/comics/dilbert/archive/images/dilbert2007048874309.gif Mar 09 11:30:21 hrw: hmm - that does answer my question somewhat ... Mar 09 11:34:13 ;D Mar 09 11:36:25 rwhitby: There is a new asterisk version that fixes a security vulnerability Mar 09 11:37:34 rwhitby: is there a problem if i clean the packages a bit.... we got something like 5 different versions + the new one Mar 09 11:37:50 rwhitby: let keep the last 3 Mar 09 11:38:10 Ifaistos: no problem here Mar 09 11:48:42 is MACHINE_TASK_PROVIDER meant to be set in distro.conf or machine.conf ? Mar 09 11:50:41 03ifaistos 07org.oe.dev * r74508525... 10/ (6 files in 3 dirs): packages/asterisk/asterisk-1.2.16.bb: Add new asterisk version that fixes a security vulnerability Mar 09 11:50:45 03ifaistos 07org.oe.dev * r5fab4cd7... 10/ (4 files in 3 dirs): packages/asterisk/asterisk_1.2.8.bb: Remove older versions of asterisk Mar 09 11:50:49 03ifaistos 07org.oe.dev * r4af36236... 10/ (5 files in 3 dirs): packages/asterisk/asterisk_1.0.9.bb: Remove old versions of asterisk Mar 09 11:50:54 03ifaistos 07org.oe.dev * r2a6cf01b... 10/ (6 files in 3 dirs): "packages/asterisk/asterisk_1.2.9.1.bb: Remove old version of asterisk Mar 09 11:51:00 03ifaistos 07org.oe.dev * red2cf8af... 10/ (6 files in 3 dirs): "packages/asterisk/asterisk_1.2.12.1.bb: Remove old versions of asterisk Mar 09 11:54:45 * rwhitby sets MACHINE_TASK_PROVIDER to task-mokogateway-everything in distro/mokoslug.conf ... Mar 09 11:56:35 rwhitby: you plan to run openmoko on the slug ? Mar 09 11:57:08 no, an openmoko gateway Mar 09 11:57:20 (companion server) Mar 09 12:01:48 morning all Mar 09 12:02:15 half of my kde session disappeared.. Mar 09 12:02:31 RP: any comments on bogofeed creation? Mar 09 12:02:34 ~curse angstrom Mar 09 12:02:39 May the fleas of a thousand camels infest your most sensitive regions, angstrom ! Mar 09 12:02:46 rwhitby: I replied to the mail Mar 09 12:03:35 aha. I also think there should be a flag. Mar 09 12:04:05 Does someone just need to re-push CoreDump|afk's changeset? Mar 09 12:04:41 rwhitby: I don't see your mail yet Mar 09 12:05:11 rwhitby: Lets wait until mickeyl gets back, I will then raise the issue with the core devs for what amounts to a vote Mar 09 12:05:51 CoreDump|afk: http://pastebin.ca/387418 <- a bogofeed implementation, to stir RP into action in putting the flag back in rootfs_ipk ;-) Mar 09 12:06:17 rwhitby: I'm waiting for mickeyl to return, I have not forgotten Mar 09 12:06:24 (since the bogofeed alternative is so ugly) Mar 09 12:06:26 rwhitby: I will look into replacing rsplit with a py2.3 function Mar 09 12:06:33 ~praise zecke_ Mar 09 12:06:35 All hail zecke_! Mar 09 12:06:41 hrw: you work at o-hand, KDE might disappear, GNOME will reappear Mar 09 12:06:51 zecke_: haha Mar 09 12:06:58 :) Mar 09 12:07:11 * RP is typing this on a remote KDE desktop... Mar 09 12:07:12 RP: I know, I was just having a little fun :-) Thanks for moderating this. Mar 09 12:07:18 zecke_: and I think about it more like 'I work for o-hand' instead of 'at o-hand' Mar 09 12:08:54 Can someone advise me how to change http://pastebin.ca/387477 to emit a task-mokogateway-everything package so I can set MACHINE_TASK_PROVIDER to that? Mar 09 12:10:11 rwhitby: You'll need to configure the -all package with the variables set for the split packages Mar 09 12:10:26 rwhitby: look at my task-base rewrite please - most of that pastebin will disappear Mar 09 12:10:47 Hey guys. Good morning. Back from .pt were I perfected my tan. Mar 09 12:10:52 hrw: got a quick pointer to save me searching my email? Mar 09 12:11:09 rwhitby: ALLOW_EMPTY_sometask = "1" probably Mar 09 12:11:15 What is our stand on a MAINTAINER field, optional or forbidden? Mar 09 12:11:23 rwhitby: "RFC task-base rewrite" Mar 09 12:11:35 Laibsch: Use Maintainers file instead Mar 09 12:11:49 RP: I have ALLOW_EMPTY = "1" at the top of that file Mar 09 12:12:05 rwhitby: You have to set it per package Mar 09 12:13:02 rwhitby: Hang on, sorry. Your package has to be listed in PACKAGES to be generated Mar 09 12:14:36 rwhitby: := is a bad idea too Mar 09 12:15:58 how about this: http://pastebin.ca/387488 Mar 09 12:17:30 hrw: Since this all has to fit in 8MB, how will I exclude some of the things that you have in the task-base rewrite? Mar 09 12:18:22 (6MB for the jffs2 rootfs, actually) Mar 09 12:18:52 rwhitby: task-mokogateway-bluetooth = task-base-bluetooth Mar 09 12:19:30 rwhitby: Yes, that should work better? Does that still not work? Mar 09 12:20:01 therein lies the problem. someone comes along and adds another package to task-base-bluetooth, and suddenly (without me doing anything, and with no way for the other person to reasonably predict it) my mokoslug distro breaks. Mar 09 12:20:13 RP: yes, that one works - was just looking for stylistic advice. Mar 09 12:21:03 rwhitby: I'd probably put RDEPENDS_task-mokogateway-everything into some variable like BASE_PACKAGES but other than that its fine Mar 09 12:21:23 rwhitby: and you use task-base which is even larger.. Mar 09 12:21:40 rwhitby: If that is a major concern, you have to fork task-base... Mar 09 12:21:46 rwhitby: task-base-minimal + task-base-bluetooth + task-usbnet-support Mar 09 12:21:51 hrw: this is first cut - I will be paring down to absolute minimum later. Mar 09 12:22:14 My question is a general one, not about this specific version of the file. Mar 09 12:22:25 (and the general question still stands) Mar 09 12:22:52 rwhitby: In general, you'd have to have a machine/distro specific task-base Mar 09 12:23:22 yes, which sort of nullifies the reason for tasks in the first place ... Mar 09 12:23:47 I have to find a happy medium somewhere ... Mar 09 12:23:49 florian: did you push your padding change yet? Mar 09 12:24:16 rwhitby: No, it does not. Before, everyone was forced to use task-bootstrap, now we have a choice Mar 09 12:24:32 rwhitby: It doesn't have to be coded into the machine either Mar 09 12:24:39 RP: true, I agree that it is much better than before. Mar 09 12:25:18 rwhitby: It is hoped we won't need too many of them and it is supposed to be a m,inimal setup Mar 09 12:25:21 I guess we could litter the task-base with ROOT_FLASH_SIZE checks ... Mar 09 12:25:43 have a prioritised lists of things that are in task-base-minimal Mar 09 12:26:03 rwhitby: Distro maintainers are still going to want to control what goes in... Mar 09 12:26:30 e.g. task-base-core-default has setserial, which is of no use on an nslu2. Mar 09 12:27:01 (not that that's a big package, but you get the general idea) Mar 09 12:27:55 rwhitby: Work out what you want in your rootfs by cloning it, we can then talk about whether its possible to share anything or not Mar 09 12:28:16 rwhitby: and if you can then look at my proposal Mar 09 12:28:43 right, hrw's proposal is along the lines of how to share some of it Mar 09 12:28:56 RP: I'll be using what's there, until things are too big, and I won't be looking at cutting it down until after hrw's task-base rewrite goes in. Mar 09 12:29:32 rwhitby: Yes, people making changes to common code might impact things. We'll just have to agree a policy for changing that common code... Mar 09 12:30:13 RP: I don't think it will be a big problem. I think most of the solution has been achieved here right now with awareness of the problem :-) Mar 09 12:31:46 why does task-base.bb have a task-base-oh-minimal (with only RDEPENDS) and a task-base-minimal (with only RRECOMMENDS) ? Mar 09 12:32:16 rwhitby: That was an example of a custom task-base with cut down depends Mar 09 12:32:36 so task-base-minimal does not RDEPEND on the kernel? Mar 09 12:32:58 Perhaps not, I don't remember the details :) Mar 09 12:33:20 task-base came from poky and I chose to leave that code in as a kind of hint/example Mar 09 12:33:42 an example of how to confuse me ;-) Mar 09 12:34:01 rwhitby: Make you think ;-) Mar 09 12:34:18 the position of those two statements in the current task-base.bb make it look like a typo. Mar 09 12:34:54 (i.e. every other task-base-something has both RDEPENDS and RRECOMMENDS, but those two only have one each, and the opposite one for each. Mar 09 12:36:03 why isn't there a RDEPENDS_task-base-minimal = "${MACHINE_ESSENTIAL_EXTRA_RDEPENDS}" in that file at least? Mar 09 12:36:22 rwhitby: It is a typo, a -oh is missing Mar 09 12:36:52 so task-base-minimal is not meant to be defined *anywhere* ? Mar 09 12:37:11 rwhitby: s/task-base-minimal/task-base-oh-minimal/ Mar 09 12:37:25 what about in PACKAGES at the top of the file? Mar 09 12:37:38 That's meant to say what it says Mar 09 12:38:08 so where is task-base-minimal defined then? Mar 09 12:38:31 (if the RRECOMMENDS_task-base-minimal in that file is meant to be task-base-oh-minimal) Mar 09 12:38:47 rwhitby: Why does task-base-minimal need to be defined? Mar 09 12:39:34 I guess it doesn't, but why have it as a package if there is no way to put stuff in the package? Mar 09 12:40:21 rwhitby: The top of the file says task-base-oh-minimal, not task-base-minimal Mar 09 12:40:56 hmm - in my task-base.bb is says both task-base-minimal and task-base-oh-minimal Mar 09 12:41:03 * rwhitby mtn pulls Mar 09 12:41:53 rwhitby: The idea was the distro/machine could say MACHINE_TASK_PROVIDER = "task-base-oh.minimal" and then get a reduced size file system Mar 09 12:42:32 rwhitby: Someone has fixed my typo incorrectly. It didn't say that in poky where the file came from Mar 09 12:43:01 ok, so http://www.openembedded.org/repo/org.openembedded.dev/packages/tasks/task-base.bb should not have task-base-minimal in PACKAGES then. Mar 09 12:43:24 rwhitby: no Mar 09 12:43:36 * RP fixes Mar 09 12:43:44 why don't we just s/-oh//g in that file completely then? Mar 09 12:45:00 RP: thanks for your patience. as you said, you succeeded in making me think with that hint/example ... Mar 09 12:45:32 rwhitby: Sorry about the typo, someone has tried to fix it and fixed it the wrong way... Mar 09 12:45:56 RP: no need to be sorry. I just wanted to make sure my understanding was correct. Mar 09 12:45:56 rwhitby: I committed it with the -oh left in, you could equally remove the -oh entirely... Mar 09 12:46:14 03rpurdie 07org.oe.dev * r31e524cb... 10/ (1 packages/tasks/task-base.bb): task-base: Fix minimal task confusion (stems from a typo in the original) Mar 09 12:46:29 rwhitby: poky has the version without task-minimal in PACKAGES... Mar 09 12:46:54 rwhitby: http://svn.o-hand.com/view/poky/trunk/meta/packages/tasks/task-base.bb?rev=836&view=auto Mar 09 12:47:06 nod Mar 09 12:47:49 should there be a task-base-minimal added, which only has MACHINE_ESSENTIAL_* in it? Mar 09 12:48:01 * RP fixes poky too... Mar 09 12:48:51 rwhitby: You could create such a version. This was why it was marked as -oh, it was someone (openedhand's) specific version Mar 09 12:49:15 rwhitby: It was left in, purely as an example Mar 09 12:49:17 I guess it's basically that and the kernel in hrw's rewrite version. Mar 09 12:49:26 (which makes sense) Mar 09 12:54:45 6029312 MokoSlug-bootstrap-image-1.0-alpha-nslu2le.rootfs.jffs2 Mar 09 12:55:09 ah, removing the kernel image will reduce that somewhat. Mar 09 12:55:26 RP: when's the kernel image removal stuff going in? Mar 09 12:58:47 * rwhitby uses IMAGE_PREPROCESS_COMMAND_append_nslu2 = "rm ${IMAGE_ROOTFS}/boot/zImage*;" Mar 09 12:58:48 in mokoslug.conf for the meantime ... Mar 09 12:59:02 CoreDump|afk: yes Mar 09 13:03:24 rwhitby: I need to write the proposal/RFC for that... Mar 09 13:07:00 zaurus in OE suxx Mar 09 13:07:16 no workiing release for it and no time for making such Mar 09 13:08:29 looks like I will really need to buy nslu2 as cheapest arm device Mar 09 13:11:15 ls Mar 09 13:11:26 oops Mar 09 13:11:32 or I will buy rs-mmc card and test soft on nokia770.. Mar 09 13:13:27 hrw, test on 770 :) Mar 09 13:13:37 I want OE to make images for 770 Mar 09 13:22:42 I want 48h day Mar 09 13:22:51 * RP too Mar 09 13:23:22 hrw: Why not just stop sleeping? ;) Mar 09 13:23:29 * Crofton too Mar 09 13:23:49 oe building for 770 would be "nice" Mar 09 13:24:00 but not "required" in my universe Mar 09 13:24:01 stefan_schmidt: sleeping is one, having time for fiancee is another Mar 09 13:24:16 finacee's are over rated :) Mar 09 13:24:32 Crofton: 770 as device is nice. but current OE distros cant provide such nice integration of net like it has Mar 09 13:24:39 yeah Mar 09 13:24:53 hrw: Some problem here with GF. Mar 09 13:25:00 wife here Mar 09 13:26:34 * stefan_schmidt wonders how all the people come to the 'single nerd' idea. Mar 09 13:35:55 03nail 07org.oe.dev * r5a015d1c... 10/ (5 files in 4 dirs): libx11: Fix floating-point error on x86_64 buildhost, closes bug 1940 Mar 09 13:45:10 hi hrw Mar 09 13:45:46 hi greentux Mar 09 13:46:15 hrw: some positiv changes on your bank account? Mar 09 13:47:41 need to check Mar 09 13:48:16 greentux: not yet Mar 09 13:48:35 hrw: long way from germany to poland... Mar 09 13:48:56 yes.. Mar 09 13:49:12 greentux: If you are giving away money I can give you my bank account number too :-) Mar 09 13:49:28 to bribe those custom officials into giving me my neo Mar 09 13:49:40 koen|away: re: tiny filesystems. No, it has to be r/w so cramfs/romfs/squashfs unfortunately won't do it. Mar 09 13:53:25 morning Mar 09 14:01:12 If I have a zd1211-firmware package, and I want it to be installed whenever the kernel-module-zd1211rw package is installed, how do I do that? Mar 09 14:04:53 03rwhitby 07org.oe.dev * rd78faded... 10/ (1 packages/zd1211/zd1211-firmware_1.3.bb): zd1211-firmware: firmware to use with the in-kernel zd1211rw driver Mar 09 14:05:25 RDEPENDS_kernel-module-zd1211rw += "zd1211-firmware" in kernel recipe? Mar 09 14:09:05 hrw: thx Mar 09 14:16:05 03rwhitby 07org.oe.dev * r4b05cfea... 10/ (1 packages/meta/slugos-packages.bb): slugos-packages: added zd1211-firmware Mar 09 14:16:12 03rwhitby 07org.oe.dev * rc63285d2... 10/ (3 files in 2 dirs): ixp4xx-kernel: Added zd1211rw module dependency on zd1211-firmware package Mar 09 14:17:02 that should be enough for zd1211 usb wifi dongles to start working on NSLU2 and kindred. Mar 09 14:25:32 I'm stuck with that error: ERROR: function do_rootfs failed. Have you ever heard of that ? Mar 09 14:25:57 03rwhitby 07org.oe.dev * r794352bd... 10/ (1 packages/tasks/task-mokogateway.bb): task-mokogateway: Initial task for an OpenMoko companion server. Designed to fit in flash on an NSLU2, or on larger devices with more flash. Will need to be updated after task-base rewrite is committed. Mar 09 14:26:02 03rwhitby 07org.oe.dev * r8db34d3c... 10/ (1 conf/distro/mokoslug.conf): mokoslug.conf: Updated to use task-mokogateway. Mar 09 14:26:26 pierrelux: look in the log file that it points you to Mar 09 14:29:55 rwhitby: it shows some error related to tar (wildcard) as well as Cannot create directory `/usr/lib/ipkg': Permission denied. I solved the first by installing bsdtar. Mar 09 14:30:27 the second one is a known harmless error in ipkg Mar 09 14:31:13 harmless but the process stops just after... Mar 09 14:31:45 not here, or on the autobuilder. Mar 09 14:32:03 then it goes with Cannot find package kernel, Check the spelling or perhaps run 'ipkg update', Cannot find package ixp-eth. and so on Mar 09 14:32:21 that's your real problem then. Mar 09 14:33:17 I sent an email on nslu2-linux last night but it didn't show up on http://tech.groups.yahoo.com/group/nslu2-general/messages Mar 09 14:35:00 it can fetch the list though: Updated list of available packages in /home/pierre-luc/slugos-3.10-beta/openslug-nslu2.tmp/rootfs//usr/lib/ipkg/lists/oe Mar 09 14:49:32 hrw: I managed to create a fully working stylus from the other broken stylus I received when I bought my replacement collie late last year. Luckily enough the guy had broken the other end ;-) Mar 09 14:54:58 Laibsch: ;D Mar 09 14:58:54 Hi everyone. I'm having problems compiling OE. Compilation stops while processing perl-5.8.7: 'You haven't done a "make depend" yet!'. This is on a AMD64 system, running Debian unstable. Any hints? Mar 09 15:01:38 amaldo: works for me on same system Mar 09 15:02:36 hrw: What do you have set as distro? I'm not sure if it should make a difference. I get it while trying to generate gpe-image Mar 09 15:03:07 amaldo: angstrom-2007.1 Mar 09 15:03:40 More specifically, It dies while building perl-native-5.8.7, on the section called "Making x2p stuff" Mar 09 15:03:54 amaldo: how old OE? Mar 09 15:04:49 hrw: I did an update last night. (cd org.openembedded.dev && mtn pull && mtn update). Am I missing to update something? Mar 09 15:05:29 then no idea Mar 09 15:05:45 try bitbake -crebuild perl-native Mar 09 15:06:21 hrw: I just updated, and trying what you suggested. Let's see. Mar 09 15:06:56 I'm trying to get a gpe_image of OpenZaurus 3.5.5, that should work, right? Mar 09 15:08:09 not tested at all Mar 09 15:09:46 hrw: Mmm.. alright... Does then setting distro to angstrom-2007.1 to generate an gpe_image work? Mar 09 15:10:54 should Mar 09 15:11:31 hrw: thanks, I will try that know. Wish me luck. Mar 09 15:28:09 Hello. Mar 09 15:28:48 hi sirfred Mar 09 15:29:09 hrw: Hi! Still no new from your seller? Mar 09 15:29:44 sirfred: yep ;( just mailed him again Mar 09 15:30:55 Is there any know reason for libsdl-x11 not being updated from 1.2.9 in oe ? Just lack of interest? Mar 09 15:31:54 probably Mar 09 15:31:59 1.2.10 and higher promises screen resolution changing using randr. I'm trying to compile under OE 1.2.11, but I'm getting into trouble (I'm just a poor autotools iliterate guy) Mar 09 15:32:36 Now that Ximageon is ready to support resolution change, It would be nice for games. Mar 09 15:32:45 yes Mar 09 15:32:59 Well, I'll keep trying. Mar 09 15:35:20 hrw: Thanks for letting me know about --set-default. Mar 09 15:35:59 hrw: And git (recent versions at least) does have progress meters for network transfers. Mar 09 15:37:26 hrw: And sure you can pull from multipe servers. But monotone, like git, has cryptographic version naming, so it should be possible to determine that the local revision and the remote revision are identical by exchanging very little information. Mar 09 15:37:29 cworth: after getting annoyed too much with git I use cogito for it Mar 09 15:37:43 cworth: monotone allow to have multiple heads Mar 09 15:38:02 hrw: Sure. So does git/cogito. How does that change anything? Mar 09 15:38:21 hrw: All that should be necessary is to exchange the crytographic name of each head. Mar 09 15:38:43 cworth: maybe. but I'm not proper person to decide ;D Mar 09 15:38:57 But here I'm waiting 30 seconds, (maybe my connection/laptop is just slower than yours) to count up stuff like this, (and no indicator or how much it's going to do): Mar 09 15:38:59 mtn: 38,346 | 47 | 12,640 Mar 09 15:39:32 Err... that's: Mar 09 15:39:34 mtn: certificates | keys | revisions Mar 09 15:39:34 mtn: 38,346 | 47 | 12,640 Mar 09 15:40:14 cworth: ask monotone guys on oftc ircnetwork why Mar 09 15:40:33 hrw: Sure. Mar 09 15:40:46 cworth: which version of mtn are you using? Mar 09 15:40:59 hvontres|home: 0.33 Mar 09 15:44:28 cworth: hmmm...well I guess it's been a while since I pulled...104 revs to go...:( plus i'm bitbraking glibc in the background Mar 09 15:47:42 cworth: I think one of the reasons the pulls sometimes seem very slow is that the db gets updated after each rev... Mar 09 15:50:48 hvontres|poodle: I don't follow. I've only been timing the "no changes" pull. Mar 09 15:51:50 cworth: oh. well, I ususally pull once a day to once every couple of days. Some devs will have a cron job pulling every couple of hours. Mar 09 15:52:12 cworth: so I know I will have changes when I pull. Mar 09 15:52:22 hvontres|poodle: And what would you guess is a typical time that a pull takes for you? Mar 09 15:53:33 cworth: 10-30 minutes probably. I usually start a pull and walk away. But I think that is partailly due to my antique build machine. Mar 09 15:54:11 hvontres|poodle: Ouch! I think we really need to see the git comparison here... Mar 09 15:54:52 And as for "faster != better", long pull times have real impacts on the ability of people to collaborate effectively with each other. Mar 09 15:55:31 * mwester thinks "Ouch!" hardly begins to describe the pain of seeing a 2.8GHz x86 box with 1GByte RAM nailed solid at 100% CPU consumption for several *hours* while doing a 200 rev mtn update... Mar 09 15:55:45 cworth: my average pull is probably 50-100 revs, so I am probably not a good benchmark here. Mar 09 15:56:01 hvontres|poodle: That's a tiny number though. Mar 09 15:56:30 * mwester and wondering what on earth (short of malware) mtn could *possibly* be doing that could take such system resources for so little work... Mar 09 15:57:23 Here, let me try pulling the last 500 revs. of cairo through git, (not that the commits are necessarily similar in scope to openembedded's, but just as a point of reference). Mar 09 15:58:11 I did a mtn db check on that same machine a couple of days ago. It took 26 HOURS to check a 126MByte monotone db. say what? how does that happen? Mar 09 15:59:11 * mwester will be watching with interest to see if others are experiencing the same mtn overhead as he Mar 09 16:04:10 cworth: ok, so a 104 rev pull with bitbake in the bacground took 17 minuts here. Mar 09 16:08:45 mwester: I think merges really kill pulls if I recall Mar 09 16:10:07 mwester: You wait hours for 200 revs? I think my avg is ~3 secs/rev Mar 09 16:10:42 OK. Pulling 500 revisions of cairo was as follows: Mar 09 16:10:44 real 0m25.061s Mar 09 16:10:44 user 0m13.694s Mar 09 16:10:44 sys 0m1.507s Mar 09 16:11:13 0.05 secs/rev Mar 09 16:11:39 (Though, again, these revisions might be tiny compared to openembedded's or something---I hope to have an apples to apples comparison soon.) Mar 09 16:12:09 Ok, I'm trying to compile with distro set to Angstrom-2007.1, but the compilation of glibc stops. Is it connected to the version problem mentioned in the wiki? Mar 09 16:12:58 I am pulling a rev every couple of seconds atm Mar 09 16:13:03 amaldo: what errors are you getting Mar 09 16:13:29 cworth: how well does git handle 3-way merges? Mar 09 16:13:48 hvontres|home: this: Preferred version 2.5 of glibc is not availabe. Then it starts building glibc-intermediate-2.5r4, and fails with this: configure: error: compiler support for __thread is required Mar 09 16:14:08 hvontres|poodle: While compiling a kernel (through bitbake), I pulled 30 revs. That took 1 minute 17 seconds. About 30 secs of that was the initial negotiation/pulling from server stuff. Mar 09 16:14:36 And I'm not running on insane hardware either. AMD X64 3700+ w/1gb ram. Mar 09 16:14:38 hvontres|home: git merging has always been just fine for me. I've only gotten conflicts exactly when I expected, and they have always been as easy to resolve as expected. Mar 09 16:16:12 Meanwhile, my oe build from last night (through MokoMakefile) died with undefined references in alsa-utils-1.0.13-r0: Mar 09 16:16:26 | state.o: In function `config_bool': Mar 09 16:16:26 | /hda3/moko/build/tmp/work/armv4t-linux/alsa-utils-1.0.13-r0/alsa-utils-1.0.13/alsactl/state.c:654: undefined reference to `snd_config_get_type' Mar 09 16:16:40 ... (and many, many similar undefined references) Mar 09 16:18:12 NAiL: heh...I think I am the one with "insane" hw....:) Mar 09 16:19:00 what hardware are we talking about? ;) Mar 09 16:19:45 NAiL: HP Kayak Xw, 512MB RAM, dual P-II 333's ... but the price was right :) Mar 09 16:20:47 NAiL: the scarry thought is that that once was one of our 3-d Cad Stations. I talked our IS guys out of it ~a year ago.. Mar 09 16:21:24 well, I don't judge based on the hardware, I ask about the task. How difficult is it to pull a revision? How much data is involved? Even if one adds security, how much CPU is reasonable per revision? And I think something is wrong, because the only explanation for why anything with MTN can take so long with so little data to process is that MTN has algorithms that are of order n-squared or such. Mar 09 16:22:16 120 MByte database - what must one do with it that takes longer than a full day of processing? Mar 09 16:22:46 Something must be wrong with the index structure, or the algorithms to result in such extreme ratios of data to processing time. Mar 09 16:22:57 mwetser: use the source :) Mar 09 16:23:13 oprofile Mar 09 16:23:24 I'm not smart enough to understand the source. I get lost in boost. Mar 09 16:23:59 I'm just a consumer of MTN. Just like I don't need to be able to design an automobile to know that one is not operating correctly. Mar 09 16:24:08 * hvontres|poodle wonders if mtn does too Mar 09 16:24:33 mwester, unfortunately complaining about open source often results in being told to solve your own problems :) Mar 09 16:24:52 * Crofton should oprofile mtn next time he pulls Mar 09 16:24:55 I know that well. That's how I ended up working on the NLSU2 Unslung project :D Mar 09 16:30:58 * hvontres|poodle *loves* having gcc and glibc built *TWICE* to get a cross compiler Mar 09 16:33:14 Crofton : How are your powerpc assembly skills ? Mar 09 16:33:40 Ifaistos, non existant Mar 09 16:33:52 my assembly skills have atrophied Mar 09 16:34:19 * Crofton used to hand assemble 6800 assembly (yes, the number of zeros id correct) Mar 09 16:34:35 Crofton : I am lost... trying to do BE/LE conversions fast Mar 09 16:34:40 Ifaistos, I added a newer version of madwifi-ng, fixed some peoples problems, broke others Mar 09 16:34:45 hmmm Mar 09 16:34:52 hmmm Mar 09 16:35:05 isn't there a special routine for that? Mar 09 16:35:09 bytesex.h Mar 09 16:35:31 i've seen all the sex today, but not the correct bytes Mar 09 16:35:40 byteswap.h Mar 09 16:36:13 /usr/include/byteswap.h should solve your problem Mar 09 16:37:04 The thing is i have this codec that sends/receives data in LE so i need something fast to change the in/out buffers fast Mar 09 16:37:33 I think byteswap.h reduces to assembly written by someone who knws Mar 09 16:39:57 I've seen the madwifi stuff. didn't test it though Mar 09 16:40:40 Crofton: heh..I did that once too...(6800 assembly that is)...had a 6809 with a pia for a mechatronics class. It was kinda amusing watching a bunch of ME's building a breadboard computer and then trying to figure out assembly language at the same time :) Mar 09 16:41:18 expect more breakage as they are changing the wifi stack Mar 09 16:41:29 heh Mar 09 16:48:14 qemux86 build fails http://builds.hackndev.com/builds/angstrom/daily-1173398401/build-qemux86.-nodebug.log Mar 09 16:49:06 hi, all! Mar 09 16:59:17 cu Mar 09 17:06:40 Crofton: 6502 assembly and hardware for myself. Mar 09 17:08:16 slapin_nb: yes, fix packaging of glibc and send a patch :) Mar 09 17:09:21 Ifaistos: is that a hardware or software codec? **** BEGIN LOGGING AT Fri Mar 09 17:26:12 2007 Mar 09 17:44:47 can the install command be used recursively in a bitbake recipe? Mar 09 17:45:23 I need to create a package with ~86 files and I assume there must be an easier way then specifying each individually Mar 09 17:53:37 xuumbi: Use wildcards? Mar 09 17:54:03 xuumbi: Or PACKAGES and FILES for your package? Mar 09 17:54:10 Laibsch: can install handle wildcards? I thought it couldn't Mar 09 17:54:27 xuumbi: I think so. Mar 09 17:54:42 Laibsch: I'm looking into the PACKAGES and FILES documentation now actually Mar 09 17:56:05 xuumbi: good Mar 09 17:56:16 There are fotentimes more than one way to do things. Mar 09 17:56:31 Taking a look at how other packages do things is often interesting as well Mar 09 17:57:10 yeah I've been doing a lot of that, with varying success Mar 09 18:08:17 mwester, we are old Mar 09 18:08:50 does is make sense that files are being packaged up correctly, but still remain in tmp/work/.../mypackage/install/? Mar 09 18:09:45 the OE documentation seems to say files that have been packaged up with not appear in that area, which is a good way of checking to make sure nothing was left behind during packaging Mar 09 18:10:38 xuumbi: we had that, but the implementation was bad, so we temporarely disabled it Mar 09 18:10:58 koen|lisboa: ohhh, ok thanks for clearing that up, I was confused Mar 09 18:11:27 with bitbake 1.8 it would remove files before they were packaged :( Mar 09 18:11:37 uh-oh Mar 09 18:15:10 koen|lisboa: It was a bitbake 1.8 problem? Mar 09 18:16:01 RP: or 1.6 + insane, or 1.6 + packaged-staging or .... or ... Mar 09 18:16:15 I suspect anything using PACKAGE_FUNC would trip the bad behaviour Mar 09 18:16:37 koen|lisboa: 'SlugOS' seems to be missing from the list of distributions in the OE bugtracker. Is that intentional or an oversight? Mar 09 18:16:52 koen|lisboa: so we change the PACKAGEFUNCS ?= to = again Mar 09 18:16:56 koen|lisboa: how is lisboa? Mar 09 18:16:56 hillct: they have their own bugtracker iirc Mar 09 18:17:04 zecke: very nice! Mar 09 18:17:10 yeah Mar 09 18:17:28 hillct: you'd have to ask rwhitby for the details Mar 09 18:17:37 but I figured it'dd be listed regardless since packages tested on slugos need to go into OE on occasion Mar 09 18:17:39 * RP back later Mar 09 18:17:43 will do Mar 09 18:21:39 koen|lisboa, you in Brazil? Mar 09 18:22:52 Crofton|home: not yet Mar 09 18:23:12 sucky transfer, so I'm spending a night in lisbao Mar 09 18:23:18 ah Mar 09 18:23:24 which is an awesome city, btw Mar 09 18:23:28 afternoon Mar 09 18:23:37 practice your portugeuse Mar 09 18:23:38 midnight Mar 09 18:23:55 australia? Mar 09 18:24:02 koen|lisboa: You are in .pt now? Just missed you then. we left .pt yesterday. Mar 09 18:24:20 anyway, does anyone know if the ep93xx stuff has touchscreen support? Mar 09 18:24:28 * natezg looks koens way Mar 09 18:24:54 natezg: none of the ep93xx people I know have a touchscreen attached to their boards Mar 09 18:24:59 hmmm Mar 09 18:25:08 gremlin[it] might have a ts Mar 09 18:25:39 koen:he doesn't seem to be on, I'll keep trollin Mar 09 18:25:44 another question Mar 09 18:25:51 koen|lisboa: if you would be so kind, please checkin this recipe: http://bugs.openembedded.org/show_bug.cgi?id=1958 Mar 09 18:26:08 has anyone seen a patch for the SPI on the ep93xx? Mar 09 18:32:21 03florian 07org.oe.dev * rfff6cc2d... 10/ (1 packages/linux/linux-ezx_2.6.16.13.bb): linux-ezx: Add patch to improve touchscreen behaviour. Mar 09 18:32:52 bd2: ping Mar 09 18:32:56 err, sorry:) Mar 09 18:55:38 bbl Mar 09 19:37:53 good evening all Mar 09 19:51:06 yeah, the weekend is almost there... Mar 09 19:59:16 brb Mar 09 20:06:02 morning Mar 09 20:11:39 zecke, koen|lisboa : quick question: is it possible to set a SRCDATE inside of a .bb file or does it need to be in a .conf? Mar 09 20:12:00 hvontres|poodle: it can be set everywhere Mar 09 20:12:38 zecke: Thanks. I wanted to do a quick test and didn't want bitbake to have to re-parse .dev again :) Mar 09 20:13:17 hvontres|poodle: it will parse the SRCDATE from the .bb file after it has parsed the configuration Mar 09 20:13:27 hvontres|poodle: so the bbfile is able to reset it Mar 09 20:17:08 LaF0rge: a comment to upstart_0.3.5.bb: 'files' is always in the FILESDIR/FILESPATH, and using ${PV} instead of 0.3.5 might be handy as well Mar 09 20:18:39 morning zecke - any luck on the rsplit rewrite? Mar 09 20:19:27 rwhitby: I just stopped doing the paid work. Will look into OE now Mar 09 20:19:49 no worries, mate. Mar 09 20:19:50 rwhitby: I would just remove the 'r' Mar 09 20:20:24 rwhitby: BTW: use ubuntu :) Mar 09 20:20:55 heh. I use Etch on one machine, but I use CentOS on the other specifically to ensure that OE builds on it Mar 09 20:21:57 since it's the work machine connected to my test rig (http://www.flickr.com/photos/rwhitby/413452216/) Mar 09 20:21:59 rwhitby: out of curiosity is python2.4 not available on CentOS or simply not the default? Mar 09 20:22:15 RHEL tops out at python 2.3 Mar 09 20:22:46 and it's not obviously available on the normal 3rd-party repos like rpmforge or centos-extras (as far as I can see) Mar 09 20:23:17 And I can't really go any further than those and keep a somewhat supported distro config Mar 09 20:23:57 rwhitby: just remove the 'r' Mar 09 20:24:18 rwhitby: hehe, nice picture Mar 09 20:25:34 hillct: ping Mar 09 20:27:17 rwhitby: pong Mar 09 20:27:42 *empty office* *finally* Mar 09 20:28:33 is oe an linux distro or a toolset? Mar 09 20:28:48 rwhitby: I was asking about having 'slugos' added to the list of distributions in OE bugzilla for packages somehow related to slugos but relevent to OE in general Mar 09 20:29:39 hillct: sure. I don't know who you should ask - I don't have access to that config. Mar 09 20:30:45 03freyther 07org.oe.dev * ra8bda210... 10/ (1 classes/insane.bbclass): Mar 09 20:30:45 classes/insane.bbclass: Use split and make it work with python2.3 again Mar 09 20:30:45 rsplit was introduced in python 2.4 and is like split but starts to Mar 09 20:30:45 split from the right side. The only difference is obviously if you Mar 09 20:30:45 have maxsplit set (which we don't). So it is fine to remove the 'r' Mar 09 20:33:55 rwhitby: the other thing is I added a spandsp library recipe that needs to get checked in and has been tested on slugosBE-3.10-beta Mar 09 20:34:20 zecke: thx muchly Mar 09 20:35:47 hillct: any reason why you set the PV and PSUBV directly, and just didn't use the date? Mar 09 20:36:53 because the sub-version is in fact pre27 Mar 09 20:37:15 and we should eventually move to use the pre-release snapshots Mar 09 20:37:21 you can set that via the .bb filename Mar 09 20:37:24 rwhitby: it should work, now let us find the next python2.4ism Mar 09 20:38:00 zecke: would only be in the new qa stuff - distros without qa still build fine. Mar 09 20:38:11 the dated image tarball was used because it incorporated patches that facilitated cross compilation Mar 09 20:38:38 black_13, a toolset to build distros Mar 09 20:38:41 rwhitby: admitingly I use python2.5 as default, so there might be more issues like this Mar 09 20:39:21 hillct: what name did you give to the .bb file? Mar 09 20:39:51 checking Mar 09 20:40:42 spandsp_0.0.3pre27.bb Mar 09 20:42:38 which reflects the naming convention of the standard upstream tarballs Mar 09 20:42:40 http://www.soft-switch.org/downloads/spandsp/ Mar 09 20:43:19 looks like pre28 should have the cross-compile fixes within the standard stream, so it may be worthwhile to use that Mar 09 20:43:34 hmm - that won't work properly when 0.0.3 is released Mar 09 20:43:53 * rwhitby wishes people just used a single version number and monotonically incremented it ... Mar 09 20:44:21 yeah Mar 09 20:44:30 thus the manual setting Mar 09 20:44:56 hillct: I'm going to change it to use a date code to avoid that. Mar 09 20:45:03 Crofton__, is there existing distro that uses oe Mar 09 20:45:11 (maybe) Mar 09 20:45:22 rwhitby: whatever you think is best Mar 09 20:45:30 black_13, ls conf/distro Mar 09 20:45:37 several distros Mar 09 20:45:51 hillct: don't know which is best yet :-( Mar 09 20:46:17 Crofton__, ill do some reading Mar 09 20:46:21 thanks Mar 09 20:47:14 how about "0.0.2+20070123" ? Mar 09 20:53:06 I saw that convention used for the speex package. works for me Mar 09 20:53:28 I guess whatever is mathmatically smaller than 0.0.3 right? Mar 09 20:54:57 2/nick hrw Mar 09 20:55:01 re Mar 09 20:55:10 0.0.2+0.0.3pre27 Mar 09 20:55:23 this is OE, Debian way Mar 09 21:18:28 NOTE: package spandsp-0.0.2+0.0.3pre27: completed Mar 09 21:22:51 03rwhitby 07org.oe.dev * rf54013c7... 10/ (3 files in 3 dirs): spandsp: A library of many DSP functions for telephony. Mar 09 21:24:30 3 files? oh, mt-attrs I guess. Mar 09 21:39:14 hi everyone, Im a newbie interested in embedded devices. I have some very lame doubts that Ive found no answer on the web. First, If I want to compile a kernel for an arm pda, should I just use gcc-arm and compile it or does bitbake handle kernel building? or am I completely wrong? Mar 09 21:39:53 work_donato: you can build kernel when you have arm toolchain Mar 09 21:40:19 So I should use oe to setup an arm toolchain? Mar 09 21:40:25 work_donato: to build software for device I suggest OE Mar 09 21:40:33 I thought oe was an arm toolchain Mar 09 21:40:35 work_donato: you can. Mar 09 21:40:58 OE is a tool to build toolchains, distros, packages, apps, libs, bootloaders, everything, order pizza Mar 09 21:41:17 so considering I just got through the "getting started" on oe's website Mar 09 21:41:23 bitbake extra larges Meat lover with extra mushrooms Mar 09 21:41:33 hrw: cool it work ^^^^^ Mar 09 21:41:41 hehe Mar 09 21:42:01 03nail 07org.oe.dev * ra3a6b606... 10/ (60 files in 15 dirs): remove openturbostation distro, add foonas Mar 09 21:42:10 how do I set it up? If it is to lame a question, can you at least point me somewhere I can get an answer? Mar 09 21:42:59 choimat: who gets the bill? Mar 09 21:43:27 hvontres|poodle: me friend /dev/null :) Mar 09 21:43:31 s/me/my Mar 09 21:43:36 work_donato: www.openembedded.org and follow GettingStarted Mar 09 21:43:38 work_donato: read gettingstarted Mar 09 21:43:49 hrw I just did, I built nano Mar 09 21:43:50 work_donato: and your ARM PDA should have linux support already Mar 09 21:44:02 hvontres|poodle: www.tinyerp.org ... I like their sandwish module ;) Mar 09 21:44:16 work_donato: bitbake virtual/kernel would build a kernel for the MACHINE you have specified... Mar 09 21:44:56 ok, does it download the kernel as well? or do I have to specify where to find it? Mar 09 21:45:41 work_donato: first: which device you have? Mar 09 21:45:45 rx3000 Mar 09 21:45:49 (rx3715) Mar 09 21:45:51 iPaq Mar 09 21:46:04 ok. supported by OE already Mar 09 21:47:18 work_donato: MACHINE="rx3000" DISTRO="angstrom-2007.1" in local.conf and then "bitbake angstrom-bootstrap-image" or "bitbake angstrom-x11-image" Mar 09 21:48:35 hrw: I think it's angstrom-console-image instead of bootstrap now Mar 09 21:48:45 a, yes. Mar 09 21:49:41 that will get me an image on the deploy dir, right? and that is the image I should boot? Mar 09 21:50:30 yes Mar 09 21:52:41 Another lame question: i want to see the source code of the kernel, do I have access to it or does bitbake hide it? Mar 09 21:53:03 meaning, can I change the source of what is to be compiled? Mar 09 21:53:32 work_donato: sure you have access to it Mar 09 21:55:33 the sources will be on the build/ directory once I compile it? Mar 09 21:55:55 right Mar 09 21:58:03 ok, that should get me started, thank you all for the attention and kindness! Mar 09 21:58:14 bye Mar 09 22:04:03 cu Mar 09 22:05:21 zecke: I have PREFERRED_VERSION_linux-rp= "2.6.17" in my distro.conf and I tried to build linux-rp-2.6.20. Now bitbake build 2.6.20 and is also building 2.6.17... Is that expeced behaviour or could this be some sort of bitbake 1.8 issue? Mar 09 22:05:57 pwd Mar 09 22:06:03 hvontres|poodle: on bitbake linux-rp-2.6.20 Mar 09 22:06:12 hvontres|poodle: it is building two kernel versions? Mar 09 22:07:12 zecke: yes. first it build 2.6.20 and now it's building 2.6.17 Mar 09 22:07:33 hvontres|poodle: on one run? funny but I don't think this is a bitbake bug Mar 09 22:07:56 zecke: Ok. It's no big deal, just kinda unexpected Mar 09 22:08:09 hvontres|poodle: someone rdepends on virtual/kernel and the deps are generated on start and linux-rp-2.6.17 is chosen to satisfy that (my pseudo explanation) Mar 09 22:08:23 hvontres|poodle: I think in a bout a hour RP returns from the pub and we can ask him Mar 09 22:08:30 zecke: ahh, that makes sense. Mar 09 22:09:05 zecke: hehe.. beer and bitbake, great combo :) Mar 09 22:09:35 zecke: I guess starting with a clean tmp dir stuff like that makes sense. Mar 09 22:11:39 03nail 07org.oe.dev * r6e3d378d... 10/ (1 conf/distro/foonas.conf): foonas: Update feed addresses to point to new domain Mar 09 22:13:09 NAiL: is foonas going to replace slugos? Mar 09 22:14:27 do you want it to? ;)' Mar 09 22:15:04 I haven't tried building it for the slug yet Mar 09 22:15:28 I'm building for the turbostation right now, will try to add support for the n2100 during the weekend. Mar 09 22:16:08 It should be capable of building for the slug, and has glibc-2.5+nptl (which was so hard to get slugos to build) Mar 09 22:17:08 There is a *lot* of work to be done with turnup/sysconf/reflash stuff to make it work on several devices. Mar 09 22:17:31 nslu2-log: you should make a task-nas (see task-mokogateway for an example), and use that in the image, and put the pack_image stuff in the machine.conf (see nslu2.conf or ixp4xx.conf for an example). then you just use bootstrap-image. Mar 09 22:18:11 NAiL: we want to replace turnup with altboot, and generalise sysconf somewhat Mar 09 22:20:01 s/nslu2-log/NAiL/ Mar 09 22:20:04 dang Mar 09 22:20:16 I was hoping nslu2-log was going to take care of the first bits :-P Mar 09 22:20:20 heh Mar 09 22:54:07 hi Mar 09 23:34:31 03rwhitby 07org.oe.dev * rcc42f592... 10/ (1 packages/tasks/task-mokogateway.bb): task-mokogateway-wifi: added zd1211-firmware Mar 09 23:43:21 time to go home for the weekend :) see y'all later :) Mar 09 23:43:58 later Mar 09 23:44:18 hvontres|poodle: I'm lost as well my Mate dealer is probably closed by now Mar 09 23:45:05 zecke: You should be able to get Mate 24/7 in Berlin, no? Mar 09 23:45:34 stefan_schmidt: probably, but not with a 2 minute foot walk Mar 09 23:45:52 zecke: true Mar 09 23:47:19 stefan_schmidt: I musn't drink mate during office hours. As Ex-CTO of convergence my boss can't stand seeing empty mate bottles :} Mar 09 23:48:19 zecke: heh Mar 09 23:50:18 I have a somewhat strange case in a bitbake recipe. I have to pull audio files and a python based conversion script which depends upon Sox to perform audio translation before packaging the files. Is there a 'correct' method for specifying this rather strange build-only dependency which only relates to a build-tool rather than anything runtime? Mar 09 23:50:50 hillct: depend on sox-native and create it if necessary? Mar 09 23:51:08 hillct: DEPENDS is for buildtime Mar 09 23:51:29 ah. that's what -native means. I thought that meant native to the target platform Mar 09 23:51:41 makes sense Mar 09 23:51:52 so in my case sox-native and python-native Mar 09 23:51:57 er Mar 09 23:52:05 bitake is python isn't it? Mar 09 23:53:17 hillct: yes, we do have a python-native as well, but you can rely on python2.3 beeing installed Mar 09 23:53:30 K Mar 09 23:54:51 thanks. will gice this a shot Mar 10 00:00:28 'night all Mar 10 00:14:11 03coredump2 07org.oe.dev * raaa47044... 10/ (4 files in 3 dirs): dfu-util: Add dfu-util and dfu-util-native from OpenMoko svn Mar 10 00:14:15 03coredump2 07org.oe.dev * r8dfef842... 10/ (1 packages/openmoko-base/openmoko-theme-standard_svn.bb): openmoko-base: Use http, not https (synced with OM svn) Mar 10 00:32:18 03freyther 07org.oe.dev * r2aeb1754... 10/ (1 classes/seppuku.bbclass conf/documentation.conf): (log message trimmed) Mar 10 00:32:18 classes/seppuku.bbclass: Switch to HTTP post, attach the build logs instead of using comments Mar 10 00:32:18 The limit of comments is too low. Use attachments to work around this Mar 10 00:32:18 issue. Also start using HTTP post to avoid really long URLs. They lead Mar 10 00:32:18 to a pipe error. Mar 10 00:32:20 To post we use a MultiPartPoster which is licensed LGPL and can be freely Mar 10 00:32:22 downloaded. You need to put it into the PYTHONPATH Mar 10 00:43:51 <_jayakumar> i noticed that diet-x11...bb builds without libi18n.a which provides XUnsetICFocus which is needed for vnc. anyone else face something similar? Mar 10 00:45:54 03nail 07org.oe.dev * r4cebc9fa... 10/ (8 files in 5 dirs): openturbostation: Remove leftover cruft Mar 10 01:11:07 <_jayakumar> i guess i should try to stub out those functions in X18NCMSstubs.c and hope that vnc doesn't really need them Mar 10 01:12:21 03nail 07org.oe.dev * rd4bb92c9... 10/ (1 MAINTAINERS): MAINTAINERS: Update my entry Mar 10 02:11:39 03zecke123 07bitbake-1.6 * r789 10/TODO: bitbake-1.6: Make a commit to test my notify chain (we cache deps the right way) Mar 10 02:25:21 03rwhitby 07org.oe.dev * ra85dec2a... 10/ (1 packages/meta/slugos-packages.bb): slugos-packages: add task-mokogateway-everything **** ENDING LOGGING AT Sat Mar 10 02:59:56 2007