**** BEGIN LOGGING AT Wed Mar 14 03:00:00 2007 Mar 14 04:27:06 03marek.vasut 07org.oe.dev * r6a261905... 10/ (4 files in 3 dirs): opie-dagger cvs: Add patch to build against sword 1.5.9. Mar 14 07:19:44 hi Mar 14 07:35:08 morning Mar 14 07:41:20 .. has anyone experience with building armv5te thumb code and glibc-2.3.4 (vs. gcc-3.4.4 )? Mar 14 08:13:43 is the monotone.openembedded.org up and using http? I can not update checkout , and I don't know if it's a firewall problem or not Mar 14 08:15:34 <_sir_maniac> I'm able to connect Mar 14 08:26:25 _sir_maniac: thank so it's probabely not using http Mar 14 08:29:26 hrw|gone : Well it seems that feed browser finally worked, after adding a lot more packages... Mar 14 08:29:49 <_sir_maniac> keesj_:for initial communications(didn't try a pull) it uses TCP port 4691, hope that helps Mar 14 08:30:42 thanks a lot I will try to setup a datapipe Mar 14 08:30:48 hrw|gone : haven't touch it... just added packages and now i can see a much bigger section list Mar 14 09:09:57 I managed to fetch the data using an ssh tunnel perhaps somebody with wiki right can add this information to the oe wiki? http://paste-it.net/1598/raw/ Mar 14 09:28:27 what are options that i can use web browser on ARM4 ? Mar 14 09:29:55 i sow 3 browsers: dillo is very simple, gpe-mini-browser crash almost every time and minimo seems good but load very heavy Mar 14 09:30:50 morning all Mar 14 09:31:15 good morning Mar 14 09:34:55 morning Mar 14 09:35:25 lkm`: or use web from o-hand.com Mar 14 09:39:59 ade|desk: ? Mar 14 09:45:56 have a look at the web bb Mar 14 09:47:29 packages/web/web_svn.bb Mar 14 10:10:11 do people generally keep seperate build trees for development / live? Mar 14 10:10:35 ie. when you cut a release, build from scratch in a fresh tree Mar 14 10:11:52 and then switch to 'debug' mode for development work - and do all that in a seperate build tree Mar 14 10:14:13 then, say for example you have an update you want to push out to your live feed... you would switch back to the 'live' build tree (which has DISTRO_TYPE set to 'release'), rebuild the package in question and sync your deploy/ipk directory to your webserver Mar 14 10:14:22 does that sound like a sane workflow? Mar 14 10:15:34 I find myself in the position of having to maintain a live distro feed - which I have no experience of doing Mar 14 10:16:24 I'm just using the deploy/ipk as a feed at the moment (rsyncing it to my webserver whenever new stuff is ready to push out) Mar 14 10:17:03 are there any potential issues I should know about when doing this? Mar 14 10:20:25 morning Mar 14 10:27:35 morning all Mar 14 10:29:16 morning RP Mar 14 10:30:20 tkp: what nslu2-linux does is has an SVN repo, which holds a stable release of the whole OE tree. We push the stable feed from there (only the stable feed manager can commit to that svn repo). The normal monotone autobuilds go to an unstable feed continuously. Mar 14 10:31:50 good morning all Mar 14 10:32:14 rwhitby: thanks... so you effectively branch off of monotones .dev when cutting a release, and merge in new bits from head as and when you are happy with them. Mar 14 10:32:22 thats kinda how I was thinking Mar 14 10:33:07 rwhitby: do you just use deploy/ipk as is for your feeds ? Mar 14 10:38:54 tkp: well, we used to, but we're waiting for the deploy/ipk directory to be put back the way it was so that we can continue doing that. Mar 14 10:39:12 (there was a change which changed the behaviour, and broke our feed generation) Mar 14 10:39:27 rwhitby: so you are one of the ones affected by the recent changes (I'm just reading that thread now) Mar 14 10:40:06 tkp: yep, I agree with the change, but as a default option, not as a forced change. Mar 14 10:40:18 hi :) Mar 14 10:40:26 (i.e. there needs to be an option to set it back the way it was before as a single dir feed) Mar 14 10:40:52 RP: any chance to talk to mickeyl and other OE core team members about deploy/ipk feeds? Mar 14 10:41:18 rwhitby: not yet Mar 14 10:43:07 good morning OE Mar 14 10:50:47 morning Mar 14 10:52:22 is this a known issue ? http://www.pastebin.ca/394468 any idea how I could resove it ? Mar 14 10:52:48 RP: Hi! still there? ;-) Mar 14 10:53:11 psokolovsky: hi Mar 14 10:54:14 RP: regarding "ipkg-make-index screws when ipk is rebuild w/o filename change": I made a separate Packages.stamps file to track timestamps, to not overload main Packages file. Sounds good? Mar 14 10:54:49 psokolovsky: That's ok with me Mar 14 10:55:16 psokolovsky: I'd have probably added it to the Packages file for simplicity but I don't mind... Mar 14 10:57:07 RP: it's just actually already too big ;-) So I took approach like with Packages.files ;-) Mar 14 10:57:31 RP: Ok, hope to submit soon new batch of my ipkg-utils patches Mar 14 10:57:51 psokolovsky: Sounds good - I'd like to see ipkg-make-index speeded up :) Mar 14 11:17:05 03slapinid 07org.oe.dev * r085e0a85... 10/ (1 packages/linux/linux-hackndev-2.6/palmld/defconfig): linux-hackndev-2.6: Update palmld defconfig for 2.6.20. Mar 14 11:21:26 btw, is there a way to reduce the size of the Packages file? clocking in at 3.5mb, it's a bit large. Mar 14 11:22:51 NAiL, there was much talk that it needs postpressing for real feed anyway... Mar 14 11:23:08 yeah Mar 14 11:23:47 03pfalcon 07org.oe.dev * r608ac639... 10/ (21 files in 10 dirs): Mar 14 11:23:47 handhelds-sa-2.4.19, handhelds-pxa-2.4.19: Remove override dirs for the versions Mar 14 11:23:47 which were removed previously. Mar 14 11:24:27 For some reason though, the packages lists end up in two places in flash on my turbostation. /usr/lib/ipkg and /var/lib/ipkg Mar 14 11:26:06 7mb of textfiles in a 13mb flash fs is a bit much :-P Mar 14 11:26:06 NAiL: grep -v ^Source: Packages >tmp;mv tmp Packages - etc Mar 14 11:26:20 huh? Mar 14 11:28:06 NAiL: remove Source: lines from Pacakges make it 50% of size nearly Mar 14 11:28:10 folks, anybody seen this '...gcc/specs: Invalid argument' weirdness when compiling meta-sdk on x86_64? It happens when compiling gcc-cross-sdk-4.1.1. Mar 14 11:33:59 RP: stupid question: are qemu/zaurus patches available? Mar 14 11:34:54 psokolovsky: I think so although I'm not sure where offhand. Were they posed on the qemu list? Mar 14 11:35:42 RP: yes, ine form of generic PXA patch, w/o specific machine definitions ;-). ok, I'll ask balrog-kun Mar 14 11:36:31 psokolovsky: ok. They do exist ;-) Mar 14 11:36:47 ok, thanks Mar 14 11:39:54 good morning all Mar 14 11:40:00 morn koen|galinhas Mar 14 11:41:17 RP: your patch is missing a -ALLOW_EMPTY_kernel-image = "1" Mar 14 11:45:47 mickey|breakfast: we got another Tosa offer Mar 14 11:46:22 really? cool, accept it. we might have a new usage for tosa as openmoko device thanks to the compatible screen :) Mar 14 11:46:34 breakfast? Mar 14 11:46:42 are you in my timezone as well? Mar 14 11:46:46 heh Mar 14 11:46:47 12:46 Mar 14 11:46:55 i'm sleeping a bit longer these days Mar 14 11:46:55 8:46 :) Mar 14 11:47:00 hehe Mar 14 11:47:09 the sun wakes me up at 6, so I've been up for a while Mar 14 11:47:41 OEDEM 3 needs to be in a tropical country Mar 14 11:49:33 koen: tudos bem? Mar 14 11:49:48 mickey|breakfast: can I send this guy to you? Mar 14 11:50:12 oxo: fine, how are you? Mar 14 11:50:20 RP: I have a kernel installed into a different partition by having it installed into /tmp, and then a postinst script moves it to the correct location. Mar 14 11:50:25 bem, obrigado Mar 14 11:52:14 how can i add all gtk puxel buff formats ? web seems that require additional libraries instead of bootstrap-image install Mar 14 11:53:03 <_law_> hi all i currently play around with alsa sound on my aximx50 could someone help with this errors? http://rafb.net/p/nhpbfN35.html Mar 14 11:53:04 lkm`: ipkg install? Mar 14 11:54:57 hrw: i want to put then into bootstrap-image. am not able to install additional packages on device Mar 14 11:55:05 oxo: http://pe360graus.globo.com/festivaldeverao/ Mar 14 11:55:44 lkm`: create own image and add any packages to PACKAGE_INSTALL in it Mar 14 11:55:49 koen: aha, uniting the duties with the pleasure, I see ;-) Mar 14 11:58:01 oxo: the festival is mandatory ;) Mar 14 11:58:15 uhuh Mar 14 11:58:32 RP: is listtaks with bitbake trunk still broken? Mar 14 12:08:30 gtk 2.101.10 seems a lot faster as 2.10.9 Mar 14 12:08:34 (on the neo) Mar 14 12:08:39 * koen|galinhas ponders Mar 14 12:09:22 zecke: Broken how? Mar 14 12:10:58 zecke: Works for me with trunk... Mar 14 12:11:31 what's the right way to pass --prefix to EXTRA_OECONF? Mar 14 12:12:07 koen|galinhas, my friend found video of the protestor. He spent several years in the south of Brazil when he was a kid Mar 14 12:12:28 He knows some Portugeuse as a result Mar 14 12:12:39 hey, a company i know is looking for a linux capable pda, and want to pay for support. anyone got any ideas? Mar 14 12:13:30 I need stuff to be installed not in /usr on target machine but somewhere else Mar 14 12:13:47 (it needswither a cf slot or have an ethernet nic builtin) Mar 14 12:19:05 anybody seen this '...gcc/specs: Invalid argument' weirdness when compiling meta-sdk on x86_64? It happens when compiling gcc-cross-sdk-4.1.1. Mar 14 12:20:27 robtaylor: ask SDG, they do linux support for a living Mar 14 12:21:00 RP: that is good, you said it is broken Mar 14 12:27:57 zecke: I think we wondered but it appears to work to me... Mar 14 12:29:13 koen|galinhas: ah, cool, i've already pointed them at SDG :) Mar 14 12:29:47 zecke: Did you get my mail about the release? Mar 14 12:35:01 RP: "Does OE have a GUI?" "We are working on it" :) Mar 14 12:35:55 koen|galinhas: Who asked that? Mar 14 12:39:24 the anti-scratchbox lobby Mar 14 12:44:48 florian: the TI dude mentioned that they have a demo for their davince and omap boards running GPE Mar 14 12:45:04 florian: I think tigrux had something to do with that :) Mar 14 12:45:27 koen|galinhas: hi Mar 14 12:45:32 koen|galinhas: heh cool :) Mar 14 12:46:00 koen|galinhas: i remember someone working for ti asking questions in #gpe some time ago. Mar 14 12:46:09 hi pH5 Mar 14 12:46:23 good morning everyone Mar 14 12:46:30 hey pH5 Mar 14 12:50:01 koen|galinhas: Tehcnically its bitbake that gets a UI ;-) Mar 14 12:52:13 chris144: i see a different problem with this one... Mar 14 12:52:37 is someone working on an ui again? Mar 14 12:52:40 robtaylor: tnb is the person to ask Mar 14 12:53:20 florian: yes... Mar 14 12:53:26 florian: go on... Mar 14 12:53:57 florian: I would be happy to fix that; if you have a hint... Mar 14 12:54:02 chris144: http://lists.linuxtogo.org/pipermail/openembedded-issues/2007-March/001715.html Mar 14 12:56:00 florian: that looks "more fixable" than the "gcc/specs: Invalid argument" thing I am seeing. Mar 14 12:56:25 i cannot find a lever to address the specs thing... Mar 14 12:56:27 chris144: indeed Mar 14 12:56:47 florian: any idea what could couse the gcc/specs: Invalid argument" issue? Mar 14 12:56:59 robtaylor: what's up... (just entered a bit ago) Mar 14 12:57:02 chris144: not really Mar 14 12:57:09 florian: ok Mar 14 12:57:36 too bad, seems to be a tough one, need gcc specialist probably Mar 14 13:21:38 florian: try to apply the last patch we did for gcc 4.1.1 and see if it makes any difference Mar 14 13:22:28 florian: basically gcc was using config values from target (through the site files) and not from the build machine Mar 14 13:34:01 hrw: sure, point him towards me Mar 14 13:41:45 mickeyl: is there some way to test does debug board+fpc are ok? Mar 14 13:42:15 hrw: openocd Mar 14 13:42:25 hrw: debug board registers USB nodes Mar 14 13:42:35 hrw: you can see it w/ lsusb Mar 14 13:42:38 koen|galinhas: my neo does not power on Mar 14 13:43:14 tomorrow I will have another neo to test Mar 14 13:43:54 * XorA is still using debug board v1 :-) Mar 14 13:44:01 mickeyl: btw, canola is opensource now (source available in a few weeks), so the neo can have a nice media center app Mar 14 13:44:07 (based on evas) Mar 14 13:44:22 canola is SDL isn't it? Mar 14 13:44:46 it's SDL now, but will be evas Mar 14 13:44:53 very cool Mar 14 13:45:09 indt is adding proper 16bit support to efl right now Mar 14 13:45:12 later this year I will have a chat w/ rasterman... Mar 14 13:45:31 canola is that app which people use on nokias (770/n800)? Mar 14 13:45:33 (hopefully, that is) Mar 14 13:45:36 I didn't manage to speak to raster on monday :( Mar 14 13:45:43 hrw: ya, that multimedia player thingy Mar 14 13:46:07 wow, gtk+2.10.10 sure feels fast. It compiles and runs cleanly for me on a compulab-pxa270/GPE if anyone is interested Mar 14 13:46:45 apaulsen: I had the same impression on my neo1973 Mar 14 13:46:49 so it isn't a fluke Mar 14 13:47:04 drat Mar 14 13:47:15 robot101 pointed the camera at me Mar 14 13:47:18 another neo, same debug board Mar 14 13:47:24 * koen|galinhas now is on the big presentation screen Mar 14 13:47:30 wave Mar 14 13:47:38 :) Mar 14 13:48:31 * koen|galinhas gets some coffee and cake Mar 14 13:48:37 and the voice told 'this person does not listen when I am talking' Mar 14 13:48:49 hehe Mar 14 13:49:51 florian: I have some UI prototype code I'll commit sometime soon, then the real work will have to start Mar 14 13:50:10 mickeyl: something for you I have Mar 14 13:50:14 File "/usr/lib/python2.4/twisted/copyright.py", line 10, in ? Mar 14 13:50:14 from twisted import __version__ as version, version as longversion Mar 14 13:50:14 ImportError: cannot import name __version__ Mar 14 13:50:37 mickeyl: I love the omoko debug board v2. Exposing the JTAG pins in a generic way is a nice touch ;-) Mar 14 13:50:54 RP: yeah, it's a pretty cool design. Mar 14 13:51:43 hrw: __version__ is not supposed to be visible at all. patch it w/ from twisted import version, version as longversion Mar 14 13:51:43 mickeyl: on desktop it works, on Z it does not. and on desktop I do not have any __version__.py Mar 14 13:51:51 ah Mar 14 13:51:57 it's a module, not an attribute Mar 14 13:52:06 sounds like a packaging bug in setup.py then Mar 14 13:52:12 * hrw is python lamer Mar 14 13:55:35 mickeyl: after dealing with packaging python stuff my respect to you grown Mar 14 13:55:42 :) Mar 14 13:55:54 ya, it's not that trivial Mar 14 13:55:58 mickeyl: anyway I will mail about 32<>64bit problems Mar 14 13:56:05 ok Mar 14 13:56:14 hrw: do you want to package my openmoko python bindings? Mar 14 13:56:32 mickeyl: now python extensions builds only becase 32 for 32 I think Mar 14 13:56:47 zecke: first have to finish paid work. then pingwinaria conference Mar 14 13:57:21 so much neo on pingwinaria :) perhaps I should visit Mar 14 13:57:40 greentux: so much? 2 only rather Mar 14 13:57:43 zecke: I have a very rudimentary .bb recipe, but it fails because pygtk isn't installed completely here. Mar 14 13:57:53 greentux: iirc there are 3 neo devices in Poland Mar 14 13:57:57 hrw: ok, "two" are much by 50 worldwide... Mar 14 13:58:01 pH5: hehe :) Mar 14 13:58:04 greentux: s/50/36 Mar 14 13:58:08 67% of polish neo :) Mar 14 13:59:18 mickeyl: can we build python with readline support??? Mar 14 13:59:31 we already do Mar 14 13:59:34 just install python-readline Mar 14 13:59:51 python will automatically pick it up if it's there Mar 14 14:00:54 mickeyl: I mean readline support in interactive python shell Mar 14 14:01:05 me too Mar 14 14:01:08 pH5: I have a unreleased v3 of them Mar 14 14:01:58 zecke: where from? Mar 14 14:02:40 RP: oh that's cool, in which language and using which toolkit? Mar 14 14:02:44 mickeyl: next commit for you Mar 14 14:02:54 mickeyl: python bindings on "abughraib" (obviously inspired by the 22c3 names of the routers) Mar 14 14:03:07 oooh Mar 14 14:03:08 nice Mar 14 14:03:32 mickeyl: e.g. our essid is palmasola Mar 14 14:04:02 florian: So far, I have the core support for bitbake so in python and there is a tty and an ncurses version. I expect a pyGtk and pyQT versions will spring up... Mar 14 14:04:38 * florian fears python Mar 14 14:04:58 florian: its scary language Mar 14 14:05:30 florian: In the bitbake case, I am now of the opinion it makes sense. If you can talk XMLRPC, you can probably talk to the bitbake server... Mar 14 14:05:36 hrw: right, and it causes scary effects if you use it Mar 14 14:06:11 RP: might be an interesting thing to try Mar 14 14:06:12 mickeyl: rev fbd5ff3ff37128ffc8addc9ff281a91c38bc6119 waits for review Mar 14 14:06:22 k Mar 14 14:06:35 mickeyl: it suxx but I needed it Mar 14 14:07:00 03hrw 07org.oe.dev * rfbd5ff3f... 10/ (3 files in 2 dirs): python: package not packaged files into python-misc (from Poky) Mar 14 14:10:00 is there a way we can get something like "today's patches" file where all the commits get packed together in an easy to use file ? Mar 14 14:10:40 Ifaistos: why do you ask for that? due the speed of mtn? Mar 14 14:10:52 hrw: you mean this just packages the rest that's there? Mar 14 14:10:57 mickeyl: yes Mar 14 14:11:00 i see Mar 14 14:11:04 yeah, it's ugly :) Mar 14 14:11:21 mickeyl: for one project I needed few files from python-misc Mar 14 14:11:21 needs to go into create-manifest.py somehow Mar 14 14:11:23 i'll think about it Mar 14 14:11:46 i'm always for high granularity. if you tell me which files you need i'll find a way to package them properly Mar 14 14:11:57 mickeyl: remember that python-2.4.4-manifest.inc has MUCH hand editions Mar 14 14:12:10 has it??? Mar 14 14:12:14 i hope not Mar 14 14:12:24 mickeyl: I did lot of cleanup in RDEPENDS Mar 14 14:12:29 uh oh Mar 14 14:12:34 why didn't you tell me? Mar 14 14:12:44 that really needs to be done in the create-manifest.py Mar 14 14:12:44 I told Mar 14 14:13:04 hmm, then i probably catched up with that Mar 14 14:13:09 mickeyl: imagine recipe which use rootfs.bbclass to generate package Mar 14 14:13:11 we'll see when i diff Mar 14 14:13:46 zecke: no mtn speed is not the matter. it will be easier to sync with other scm's Mar 14 14:14:03 Ifaistos: use tailor? Mar 14 14:14:19 Ifaistos: mtn pull/update;svn commit -m"another sync with mtn" Mar 14 14:14:37 zecke: does not work. i have put a bug report for sometime now but nothing is moving Mar 14 14:14:47 Ifaistos: tailor worked unattended for weeks Mar 14 14:15:00 I'm trying to push out an upgrade which includes a new kernel and kernel modules Mar 14 14:15:20 but I'm finding that they don't get loaded correctly once installed Mar 14 14:15:54 I also tried to run the modutils initscript manually to sort out modules.dep etc Mar 14 14:15:59 but that doesn't work Mar 14 14:16:23 the initscript uses `depmod -Ae` which is not a valid command Mar 14 14:16:23 zecke: -> http://progetti.arstecnica.it/tailor/ticket/97 Mar 14 14:16:29 Ifaistos: some revs needed the usual str.encode/decode help Mar 14 14:16:31 there is no -A or -e Mar 14 14:17:14 actually... sorry. there is, they are just not listed in --help Mar 14 14:17:23 hrw : mtn does not like svn unfortunatelly Mar 14 14:17:32 running `depmod -a` manually seems to work Mar 14 14:17:41 but not -Ae Mar 14 14:18:47 -A - "This option scans to see if any modules are newer the modules.dep file before any work is done: if not, it silently exits rather than regenerating the files." Mar 14 14:19:28 zecke: probably will do it myself. make a cron job to do a new checkout then diff that with the prev day Mar 14 14:20:26 zecke : would it be possible to add another distro to tinderbox ? Mar 14 14:20:46 Ifaistos: sure, if you plan to report to it Mar 14 14:21:02 zecke : that's the idea Mar 14 14:21:14 Ifaistos: name? Mar 14 14:21:19 zecke : oplinux Mar 14 14:22:41 Ifaistos: http://ewi546.ewi.utwente.nl/tinderbox/showbuilds.pl?tree=oplinux Mar 14 14:22:54 Ifaistos: what is your login to bugzilla? Mar 14 14:23:27 zecke : don't remember :( Mar 14 14:23:45 Ifaistos: if you remember I can grant you editor rights for the tree Mar 14 14:23:51 zecke : koroneos@digital-opsis.com Mar 14 14:23:56 zecke : skoroneos@digital-opsis.com Mar 14 14:27:42 zecke : thanks ! Mar 14 14:28:10 Ifaistos: do you want to try seppuku as well? Mar 14 14:28:23 Ifaistos: or extend my autobuilder script? Mar 14 14:29:15 * Ifaistos hopes he does not have to perform one Mar 14 14:30:05 zecke : Haven't looked at them to tell you the truth, but i could try them out Mar 14 14:35:19 total used free shared buffers cached Mar 14 14:35:19 Mem: 2120972 2094660 26312 0 102988 1208328 Mar 14 14:35:38 no matter how much it is it will get used. Mar 14 14:38:24 hrw: IIRC, linux will fill up the unused memeory with file cache (1,208,328 K in your case) Mar 14 14:39:38 hrw: install exmap Mar 14 14:39:50 hrw: or 58% of your memory usage is cache right now Mar 14 14:41:40 * hvontres|poodle wishes he had room for 1,208,328 K of file cache on his OE machine Mar 14 14:41:55 hrw: That's standard kernel behaviour, it uses anything it can for cache as free memory is a waste ;-) Mar 14 14:42:46 I know Mar 14 14:43:12 as long as system run without breaks I'm happy. Mar 14 14:43:28 morning Mar 14 14:46:04 RP,hrw: I have been thinking about how to handle flash-limited devices like poodle and sl-c700. With 2.6 we can set them up to have / on SD but I think it would be nice if we could still have at least a console image in flash in case the SD gets borked. Would it make sense to create a new machine that includes tha base machine config and then overides things for root-on-sd? Mar 14 14:47:49 hvontres|poodle: no, we rather need to adapt nslu root-on-other-media way? Mar 14 14:48:22 hvontres|poodle: No, a new machine shouldn't be needed. I'd suggest just using something like altboot Mar 14 14:49:29 RP: One of the issues I am trying to figure out tho is how to have linux-rp build two versions of the kernel with different cmdlines Mar 14 14:50:08 RP: actually, I am working on cleating up my altboot bootloader script :) Mar 14 14:50:14 tnb: oh, i was just asking around for companies that support linux on pdas Mar 14 14:51:31 robtaylor: ah, ok Mar 14 14:51:45 robtaylor: we do. sdgsystems.com Mar 14 14:52:22 hvontres|poodle: We probably need to find a way of allowing us to change the kernel commandline at runtime... Mar 14 14:52:28 tnb: yeah, i;ve already pointed them in your direction :) Mar 14 14:52:44 RP: well, do you want to try to release 1.8.0 yourself? Mar 14 14:52:53 RP: heh...that of course would be the *best* solution :) Mar 14 14:54:16 RP: do you think somethink like this might work: http://patchwork.ozlabs.org/linuxppc/patch?id=5708 Mar 14 14:54:40 RP: uboot? Mar 14 14:55:23 hrw: No, We could add some kind of shim to the kernel though Mar 14 14:55:52 zecke: How difficult is it and am I likely to screw it up? :) Mar 14 14:56:26 RP: unlikely, but even if you would screw up we can easily recover Mar 14 14:56:56 RP: a send a random list of things to do. Mostly just check packaging (MANIFEST) and make sure to bump the version numbers Mar 14 14:57:02 hvontres|poodle: Interesting although the compressed zImage will be a pain to expand,edit and recompress again Mar 14 14:57:10 RP: and on the berlios.de site, I always enter garbage and then undo it Mar 14 14:58:17 zecke: ok, I'd like to give it a go... Mar 14 15:02:24 RP: it looks like they use objdump and dd in a scrtipt: http://patchwork.ozlabs.org/linuxppc/patch?id=5577. I guess it's intended more for modifying an existing zImage for a new machine Mar 14 15:05:52 hvontres|poodle: That doesn't cover decompression and recompression again, unless it stores the commandline in the uncompressed part of the binary? Mar 14 15:08:41 03koen 07org.oe.dev * r92449623... 10/ (1 packages/loudmouth/loudmouth_1.1.1.bb): loudmouth: pdate to 1.1.1 Mar 14 15:08:57 RP: I pimped your LZO patch to the OLPC and indt people Mar 14 15:09:45 koen|galinhas: :) It should be useful for OLPC... Mar 14 15:09:55 You didn't pimp hard enough, apparently Mar 14 15:10:23 CosmicPenguin: What makes you say that? Mar 14 15:10:32 Because this is the first I've heard of it Mar 14 15:11:29 CosmicPenguin: I posted a series to LKML and linux-mtd which boosts jffs2 filesystem reads by about 40% at a cost of slight loss of compression ratio Mar 14 15:11:51 RP: do you have average numbers? Mar 14 15:12:10 zecke: for what, the loss in compression? Mar 14 15:12:22 RP: yes, for the loss of compression Mar 14 15:12:29 Nice Mar 14 15:12:30 I know it highly depends on the files in question Mar 14 15:12:46 I think that dwmw2 is on vacation, so we might not have heard about it Mar 14 15:13:00 RP: I am not sure, but it looks like the buffer might be located in the uncompressed part of the zImage...but then I am kind of a kernel neophyte. Time to look at kexec arm I guess Mar 14 15:13:04 the brazilian dude knows about it now Mar 14 15:13:08 (marcello?) Mar 14 15:13:48 | ../src/write-mgr-file > gabble.manager| ../src/write-mgr-file: line 105: /data/build/koen/OE/build/tmp/angstrom/work/armv4t-angstrom-linux-gnueabi/telepathy-gabble-0.5.5-r0/telepathy-gabble-0.5.5/src/.libs/lt-write-mgr-file: cannot execute binary file Mar 14 15:13:49 Marcelo Tosatti Mar 14 15:13:56 * koen|galinhas goes out to kick robot101 Mar 14 15:14:53 zecke: 7% worst case for a full Linux image according to one test Mar 14 15:15:23 hvontres|poodle: kexec doesn't support commandlines yes Mar 14 15:15:25 yet Mar 14 15:15:47 CosmicPenguin: That would explain his silence about the patches... Mar 14 15:15:54 RP: I know. Any idea what would be involved in getting that to work? Mar 14 15:16:18 <||cw> I'm interested in running linux on a device I have (HP digital sender 9100C) that's ARM based but boots of a hard disk with some OS I don't recognize. any tips or sites on getting started? Mar 14 15:16:21 RP: yeah - but I do think something like that would be pretty interesting for OLPC Mar 14 15:16:29 hvontres|poodle: The commandlines are disabled in the zaurus kernel anyway so it will be tricky Mar 14 15:16:52 CosmicPenguin: Since the decompression is faster, less battery use too Mar 14 15:17:09 speed is our friend at this point Mar 14 15:18:02 CosmicPenguin: I'd like to see the LZO patches in mainline. I think they have a good chance. I might try to get them into -mm Mar 14 15:18:34 Its worth a shot - when our kernel guy gets in, I'll talk to him about getting them into the olpc tree Mar 14 15:19:00 unstable tree, that is Mar 14 15:19:27 RP: Is that a sharp bootloader thing or some other reason? Currently I have altboot setup to boot a normal kernel from flash and then let the user choose wich kernel to kexec from the SD card...It's really neat for testing.. just put your 2.5.17 and 2.6.20 kernels + modules on SD and see what works :) Mar 14 15:20:04 hvontres|poodle: The sharp bootloader passes a commandline that the kernel can't use so I disabled commandline support Mar 14 15:21:37 RP: ahh, I see. So even if we cmdline editing working, we would still need two versions of the kernel...:( Mar 14 15:22:04 ~lart SHARP's braindead bootloader Mar 14 15:22:05 * ibot resizes SHARP's braindead bootloader's terminal to 40x24 Mar 14 15:22:33 hvontres|poodle: right. Which is why I wondered about a shim to nuke the bootloaders commandline Mar 14 15:22:56 Someone with some ARM assembler knowledge will need to write it... Mar 14 15:23:15 RP: cr? :D Mar 14 15:23:17 * hvontres|poodle does not grok this use of shim. Mar 14 15:23:47 * hvontres|poodle only knows of the kind you use to make stuff fit after you made it too small... Mar 14 15:23:48 RP: is shim not already written for some of the slug devices Mar 14 15:23:58 hvontres|poodle: piece of code inserted between the bootloader and the kernel which wiped out the bootloaders commandline Mar 14 15:24:08 can you tell me if mozilla minimo support cookies ? Mar 14 15:24:18 XorA: That's a shim but with a different purpose Mar 14 15:24:25 and how can i enable them ? Mar 14 15:24:57 RP: oh, I thought it was to remove duff bootargs and replace with good, Ill have to read emails better Mar 14 15:25:20 XorA: Perhaps it was, I thought it was to switch little/big edian Mar 14 15:25:32 RP: ahhh, thanks. yet another piece of trivia to pull out later :) Mar 14 15:29:42 zecke: I've just seen the mail, its exactly what I need, thanks. I'll have try later on :) Mar 14 15:33:23 RP: It looks like arm-kernel-shim might be what we are looking for..: http://wiki.buici.com/wiki/Arm_Linux_Kernel_Shim Mar 14 15:33:28 RP: feel free to praise yourself :) Mar 14 15:33:38 RP: It's even in .dev :) Mar 14 15:39:29 zecke: I might praise the release ;-) Mar 14 15:41:38 hvontres|poodle: Its a bit bigger than I was thinking but yes, something like that should help/work Mar 14 16:09:43 bzr selftest svn... hmm feisty is broken :( Mar 14 16:13:35 can anyone explain why depmod has been passed the -A flag in the modutils initscript? Mar 14 16:14:28 this flag is supposed to check the timestamps of modules found of the system, and only update modules.dep if it finds a module newer than modules.dep Mar 14 16:15:16 but when I install a new kernel module on the target system, the kernel module seems to retain the timestamp from when it was actually built Mar 14 16:16:46 so, after upgrading my kernel from 2.6.16 to 1.6.18, modules.dep is not getting refreshed by the initscript since the new (empty) modules.dep file in the 2.6.18 tree is newer than any of the kernel modules Mar 14 16:23:28 I mean that when installing a new kernel, kernel-image creates a new modules.dep file in the new /lib/modules tree Mar 14 16:24:12 but any of the new modules that I install for the new kernel seem to retain the timestamp from when the ipk was created Mar 14 16:24:28 so modules.dep never gets updated Mar 14 16:27:23 tkp: there was discussion about it on OEML which ended in this way Mar 14 16:27:58 hrw: ended what way? with the problem I'm faced with now? Mar 14 16:31:24 * tkp can't find the thread Mar 14 16:33:20 -a was changed to -A Mar 14 16:33:51 hrw: it's -A that seems to be the cause of the trouble tho Mar 14 16:35:26 it's because modules.dep gets a fresh timestanp when the new kernel-image package is installed Mar 14 16:35:48 03hrw 07org.oe.dev * r820eb34b... 10/ (3 files in 2 dirs): openmoko-pim: tweaked RDEPENDS to same as Contacts/Dates (from Poky) Mar 14 16:35:55 which is later than the timestamp that the new kernel modules were built Mar 14 16:36:07 therefor, -A causes depmod to do nothing Mar 14 16:41:30 03hrw 07org.oe.dev * r0f247ea6... 10/ (17 files in 6 dirs): contacts, dates, tasks: moved to pimlico directory (from Poky) Mar 14 16:47:17 Is there a maintainer for JamVM/Kaffe/Classpath? Mar 14 16:48:23 apaulsen: Kaffe? It looks like no there is not. I tried to fix the related bugs ages ago. Nobody showed interest and I did not succeed. Mar 14 16:48:31 If you are interested go right ahead. Mar 14 16:48:38 Take a look at the BTS. Mar 14 16:51:59 Laibsch: I'm using JamVM + Classpath-gtk to run some customer apps and possibly a frontend for a demo app. It seems to run, but is buggy and Swing is very slow Mar 14 16:53:35 I tried to update classpath a few weeks ago, but it requires a recent gconf, which is something OE doesn't have. I may attempt to locate and remove that requirement if possible. Newer classpath is supposed to use Cairo for graphics. Mar 14 16:56:21 apaulsen: Would it be better to just get newer gconf into OE ? Mar 14 16:57:14 possibly, but it seems like the distros favor gconf-dbus which was apparently abandoned Mar 14 17:07:51 is there a means through tinylogin adduser to set up an account which is locked? seems that --disabed-password doesn't quite do it Mar 14 17:08:36 but I forget the distinction between an encrypted password of 'x' and '*'. Thought only the latter indicates locked Mar 14 17:22:44 gtk+2.8.9-r7 complaining about not able to link to Pango, is that a common issue? Mar 14 17:25:04 03nail 07org.oe.dev * r6fc204c3... 10/ (1 packages/linux/linux_2.6.20.bb): linux: Add do_deploy for n2100 Mar 14 17:25:09 03nail 07org.oe.dev * r0270b21f... 10/ (1 packages/linux/linux_2.6.20.bb): disapproval of revision '6fc204c35354ea319370965bba9aed98448036fd' Mar 14 17:25:14 03nail 07org.oe.dev * rb89e1b53... 10/ (8 files in 4 dirs): foonas-init: Remove unneeded scripts Mar 14 17:25:42 nail: oopps... Mar 14 17:26:44 yep. That -A in the modutils initscript is a real problem Mar 14 17:26:55 It stops you being able to do a kernel upgrade Mar 14 17:27:05 or I'm doing somethin wrong Mar 14 17:27:37 I'm doing it in two passes... I install kernel and kernel-image-2.6.18 and then restart Mar 14 17:28:19 after doing that, installing any other kernel modules results in modules.dep not being updated Mar 14 17:29:44 tkp: have you tried it the other way around? I rember donig that on my Zaurus a while back. I used ipkg to upgrade the kernel-modules and the flashed the new kernel. Mar 14 17:29:54 my oe experiance ends when I want to build the first package it can't download http://download.savannah.gnu.org/releases/quilt/quilt-0.45.tar.gz Mar 14 17:30:13 hvontres|poodle: no I haven't tried that... worth a shot Mar 14 17:30:32 what I dnt understand tho... Mar 14 17:30:52 a kernel module has depmod -a in it's postinst script Mar 14 17:30:59 yet this doesn't work Mar 14 17:31:08 but running depmod -a manually does Mar 14 17:33:10 also. If I run /usr/lib/ipkg/info/kernel-module-uinput.postinst manually... then modules.dep gets updated correctly... just not when it's run by ipkg Mar 14 17:52:41 can I try something more basic then nano? Mar 14 17:54:10 keesj : well it has nothing to do with nano, without quilt you can build the rest of the toolchain Mar 14 17:54:49 I think it's the website that is down Mar 14 17:58:28 can somebody help me get that file? Mar 14 18:01:11 try agian the gnu site gets unresponsvive sometime due to high load Mar 14 18:23:14 cu Monday Mar 14 19:04:47 hey people Mar 14 19:22:34 03lenehan 07org.oe.dev * rc7ea86d4... 10/ (3 files in 3 dirs): Mar 14 19:22:34 gmp 4.2.1: Allow gmp 4.2.1 to work for sh arch by apply a patch from Mar 14 19:22:34 the gmp home page which ensure the gmpn_add_nc and gmpn_sub_nc Mar 14 19:22:34 functions are included when they are required. Closes #1605. Mar 14 19:25:36 03ifaistos 07org.oe.dev * rf97f8afb... 10/ (3 files in 3 dirs): packages/jpeg/jpeg_6b.bb: Add patch that allows libjpeg to compile using libtool on varius architectures Mar 14 19:25:40 03ifaistos 07org.oe.dev * r4b73c331... 10/ (1 site/powerpc-linux): site/powerpc-linux: Add definitions for mysql and sudo Mar 14 19:25:47 03ifaistos 07org.oe.dev * r24f47376... 10/ (1 packages/perl/perl-5.8.7/config.sh-powerpc-linux): packages/perl/perl-5.8.7/config.sh-powerpc-linux: Fix typo Mar 14 19:25:52 03ifaistos 07org.oe.dev * rb6b79b88... 10/ (1 packages/gift/gift_0.11.5.bb): packages/gift/gift_0.11.5.bb: Gift required libtool in order to build, so add dependacy Mar 14 19:26:29 Good evening Mar 14 19:27:14 XorA|gone: I have to test that new mplayer, it seems a lot of progress was made. Mar 14 19:27:45 sirfred: yes, pxa overlays from Tim Chick Mar 14 19:27:59 XorA|gone: Well, and that iwmmx support, no? Mar 14 19:28:38 XorA|gone: Did you have time to test the new Ximageon release? It should provide a full randr support. Mar 14 19:29:06 XorA|gone: Also some bugs about cursor and overlay position were fixed Mar 14 19:29:36 sirfred: Ive had zero time recently Mar 14 19:30:07 XorA|gone: ok Mar 14 19:41:38 cu Mar 14 19:48:32 good evening and morning all Mar 14 19:50:15 afternoon :) Mar 14 19:51:22 JoeSchmo: ok, good afternoon for you alone :-) Mar 14 19:51:34 hehe :p Mar 14 19:52:20 But I guess America will be fully lit by afternoon in a few minutes, right? Mar 14 19:53:46 morning all Mar 14 19:57:54 rwhitby: good morning Mar 14 20:16:22 morning Mar 14 20:18:48 gm Mar 14 20:26:48 rwhitby: FYI - madwifi r2182 doesn't built for my i486 target (wrap) due to CONFIG_VLAN_8021Q=m. It's fixed in r2187 so I'm going to add that now. Mar 14 20:29:10 v8jlene: cool Mar 14 20:56:36 has anyone seen something like that -> http://rafb.net/p/t2JgQ065.html Mar 14 20:57:29 Ifaistos: nope, what bitbake is that? Mar 14 20:57:53 1.6.7 Mar 14 20:59:45 hmm Mar 14 21:00:09 * NAiL sneaks away before someone starts pointing fingers Mar 14 21:01:16 Ifaistos: such code seems to be only in bitbake-1.3.1 ... Mar 14 21:02:03 http://rafb.net/p/kGP4h391.html Mar 14 21:02:27 Ifaistos: in classes/patch.bbclass in fact Mar 14 21:02:27 cyrilRomain: which code? Mar 14 21:02:49 i removed the tinder stuff and trying to rebuild it. Mar 14 21:06:28 nop does not build Mar 14 21:07:48 Ifaistos: __builtin__.CmdError is really weird Mar 14 21:08:07 Ifaistos: http://lists.linuxtogo.org/pipermail/openembedded-issues/2006-October/000055.html Mar 14 21:08:39 Ifaistos: you don't use bitbake 1.6.7 :} Mar 14 21:08:55 hmm Mar 14 21:09:00 Ifastos, i had similar error yesterday for quilt ("did it need to be unpacked...") when trying to compile glibc. After I cleaned my build/tmp however it was ok. Mar 14 21:10:52 zecke : full output -> http://rafb.net/p/qxs96d72.html Mar 14 21:11:14 hmmm yesterday i synced my local (svn based) copy with OE Mar 14 21:12:11 Ifastos, yes i guessed that's why my glibc needed recompilation Mar 14 21:13:13 ok doing a clean build then Mar 14 21:15:36 hmm the libx11_X11R7.1-1.0.1.bb required by diet-x11_X11R7.1-1.0.1.bb already apply makekeys.diff patch Mar 14 21:16:14 same name, same content. Can it make bitbake crazy ? Mar 14 21:17:57 Ifaistos: can you try removing file://makekeys.diff;patch=1 from diet-x11_X11R7.1-1.0.1.bb ? Mar 14 21:18:23 hmmm i am clearing tmp Mar 14 21:18:35 if it fails again, i'll try that also Mar 14 21:22:18 v8jlene: thanks for your help yesterday btw. It worked. I submitted the fix in bug #1987 Mar 14 21:38:54 cyrilRomain|bbl: No problem, good to see you worked it out ;) Mar 14 22:06:05 03lenehan 07org.oe.dev * re3d3c97f... 10/ (1 packages/madwifi/madwifi-ng_r2187-20070309.bb): Mar 14 22:06:05 madwifi_ng: Add new version r2187. This version adds a single VLAN Mar 14 22:06:05 related fix to the r2182 version which is needed to fix builds for Mar 14 22:06:05 kernels with CONFIG_VLAN_8021Q=m. Mar 14 22:06:35 !oebug 1987 Mar 14 22:06:36 * * Bug 1987, Status: NEW, Created: 2007-03-13 22:55 Mar 14 22:06:37 * * c.romain(AT)laposte.net: glibc-2.5 QA issue: non debug package contains .debug directory Mar 14 22:06:38 * * http://bugs.openembedded.org/show_bug.cgi?id=1987 Mar 14 22:20:30 good night Mar 14 22:24:42 03lenehan 07org.oe.documentation * rade635fe... 10/ (1 usermanual/reference/class_autotools.xml): Mar 14 22:24:42 class_autotools: Update the description of the site files to account for the Mar 14 22:24:42 changes to enable multiple site files, package specific site files, common Mar 14 22:24:42 site files and site file aliases. Mar 14 22:28:25 I have a licensing question, anyone on that can field it? Mar 14 22:29:21 natezg: Depends on what the question is I suppose... Mar 14 22:36:00 If a binary package is generated with OE and released Mar 14 22:36:20 do all scripts used to generate it automatically fall under the GPL? Including local collections? Mar 14 22:39:50 natezg: Heh, I'd have no idea. All of the OE metadata and classes are under a MIT license, not GPL though (see copying file at top level of an OE checkout) so I wouldn't think they'd have to be released under GPL. Mar 14 22:41:38 In theory src_distribute_local should make available everything that has to be released in source form. Mar 14 22:41:48 The mailing list might be a better place for the question? Mar 14 22:42:47 'night all Mar 14 22:58:02 RP: nslu2-linux use the arm-kernel-shim to have a common ixp4xx kernel, but change machine id, ram size, cmdline, etc on different machines each of which have broken bootloaders. Mar 14 23:34:12 03lenehan 07org.oe.documentation * rc5407e38... 10/ (4 files in 3 dirs): Mar 14 23:34:12 usermanual: Add a siteinfo class section to the reference manual. Move the Mar 14 23:34:12 siteinfo class related stuff from the autotools class section into the Mar 14 23:34:12 siteuinfo class section and add details on the SITEINFO_ENDIANESS and Mar 14 23:34:12 SITEINFO_BITS variables. Mar 15 00:13:39 03lenehan 07org.oe.documentation * ra1a84461... 10/ (1 usermanual/chapters/recipes.xml): recipes: Fill out the section on DEFAULT_PREFERENCE. Mar 15 00:17:16 lennert thanks for updating **** ENDING LOGGING AT Thu Mar 15 02:59:59 2007