**** BEGIN LOGGING AT Wed May 25 23:59:56 2005 May 26 00:02:42 yes it is May 26 00:03:19 qwindowsystem_qws.cpp May 26 00:03:28 cool May 26 00:03:37 I could have sworn it wasnt there before :) May 26 00:03:56 guessing it's part of the full screen input method support? May 26 00:04:29 good morning all May 26 00:07:14 hey May 26 00:29:56 hi people May 26 00:30:25 would it be possible to build a gpe-image without letting bitbake also wanting to build a kernel? May 26 00:31:28 it should be May 26 00:32:32 I just tried "ASSUME_PROVIDED = virtual/kernel", but now packages are failing because they need some kernel files May 26 00:33:01 hmmm May 26 00:33:40 without that rule, I get that _append bug May 26 00:34:24 that should be fixed with latest svn of bitbake May 26 00:34:34 according to mickey|zzZZzz May 26 00:34:37 thanks, I'm going to try it immediately May 26 00:42:54 ~lart management May 26 00:42:54 * ibot eats management's liver with some fava beans and a nice chianti May 26 00:43:39 morning May 26 00:44:32 03koen 07 * r1.3393.2.1 10openembedded/packages/maemo/ (3 files in 2 dirs): Add outo and hildon-lgpl, part of the maemo.org suite May 26 00:46:39 koen: questing to get a 770 lookalike before it even hits market? May 26 00:46:49 yup May 26 00:47:07 the kernelsource is still missing, so no complete OE integration yet May 26 00:58:59 hey bluelightning May 26 00:59:11 hi koen May 26 00:59:26 hi all May 26 01:01:21 morning all May 26 01:02:01 morning RP May 26 01:02:26 hi RP, XorA May 26 01:02:27 hey RP May 26 01:05:24 03koen 07 * r1.3398 10openembedded/packages/maemo/libosso_0.8.9.bb: Add libosso 0.8.9, part of the maemo.org suite. Needs some packaging improvements. May 26 01:07:07 morning all May 26 01:07:16 hey ade|desk May 26 01:12:33 has the world calmed down after the announcement of the nokia device yet ?? May 26 01:12:43 nah May 26 01:13:03 oh dear May 26 01:13:15 too bad it lacks a keyboard, somewhere in there.... May 26 01:13:45 bt keyboard May 26 01:14:02 nokia makes a pretty nice one May 26 01:14:18 eh, yeah, I know, but I don't see myself lugging around the device & a keyboard :) May 26 01:14:39 770 + Z -> x2x May 26 01:15:05 koen: :-) May 26 01:15:07 is it me or do people just want the same stuff as a laptop but in a smaller form? like an a5 sub notebook akin to the toshiba libretto rather than a real pen driven pda May 26 01:15:09 hehe, now there is a solution to the problem :) May 26 01:15:36 ade|desk: My Z is a notebook replacement, so, yes, ;) May 26 01:15:42 or a sony vaio u71 May 26 01:16:48 (and my handwriting being a cross between sanscrit and stone tablet language (spijkerschrift??) no form of grafiti is going to be useful to me :) May 26 01:18:01 ade|desk: thats what I want May 26 01:18:15 Hertog: blame your teachers , sue the school, get funding for a device that will translate spoken word to type .. job done ;) May 26 01:20:50 's not going to work here I guess :) May 26 01:33:24 hmm, what was the command to force a fsck on next reboot... May 26 01:34:23 the qws documentation leaves a bit to be desired :p May 26 01:44:12 treke: email and complain :) May 26 01:46:57 hehe May 26 01:47:12 found a post on qt-interest, they dont want people using this anywyas :) May 26 01:47:14 too bad May 26 01:49:48 trying to see if a full screen IM is as easy as I suspect it will be May 26 02:13:23 03koen 07 * r1.3399 10openembedded/packages/bluez-libs/bluez-libs_2.16.bb: bluez-libs_2.16.bb: inherit pkgconfig May 26 02:20:24 setserial /dev/ttyS0 port 0x401, shouldn't that just work? or is the port way out of range? (I get 'invalid argument') May 26 02:28:30 anyone here? May 26 02:29:52 type /who in your irc client May 26 02:30:45 just joined, from #ubuntu, and felt the drop in traffic.. May 26 02:32:10 koen: is there anyone here who looks after the website? The front page needs fixing. May 26 02:33:15 and I guess the topic for this channel should also be changed. May 26 02:37:17 pschulz01_: I think kergoth or mickey|zzZZzz can change the site May 26 02:37:24 what's wrong with the topic? May 26 02:37:43 miniconf was a month ago.. May 26 02:37:49 dev May 26 02:41:06 03koen 07 * r1.3400 10openembedded/packages/matchbox-wm/matchbox-wm_0.9.5.bb: matchbox-wm_0.9.5.bb: new file May 26 02:43:24 koen: can you initiate a oe-devel -> openembedded transfer? May 26 02:44:13 only mickeyl and kergoth can afaik May 26 02:44:46 koen: ah, ok. Then we'll just wait for Kergoth to get setup in the new house I guess :) May 26 02:45:26 you can do bk parent bk://oe-devel.bkbits.net/openembedded May 26 02:45:32 just don't clone from there May 26 02:46:43 koen: from within my current openembedded repo? May 26 02:47:02 yep May 26 02:56:26 koen: thanks, works May 26 03:16:11 hi May 26 03:16:24 hey hrw May 26 03:16:25 someone know how to make automount mmc card under 2.6? May 26 03:16:59 I also have ide.agent for cf cards (based on mmc.agent) but both wont work as udev create device after they are started ;( May 26 03:18:00 I will have to push some changes from my husky to repo... /etc/init.d/devices fix for udev is already in repo iirc May 26 03:22:01 koen: we dont have automounting cards now under 2.6? (automounted after insert and 2.6 = 2.6.12-rc+) May 26 03:22:21 it works for my sd card May 26 03:22:47 hrw: something broke VERY recently, could be the issue kergoth noted where no-one can guarantee that udev and hotplug are synchronous May 26 03:23:17 hrw: apparently udev can call scripts on device creation, however last time I checked udev was still pretty much undocumented May 26 03:23:23 hrw: We need to use the new release of udev May 26 03:23:30 * mickeyl yawns May 26 03:23:35 morning mickeyl May 26 03:23:35 morning boys & girls May 26 03:23:51 'morning mickeyl May 26 03:23:53 mickeyl: Did you get the new relese of udev working? May 26 03:24:16 didn't try yet May 26 03:24:21 it compiles though May 26 03:25:26 I'll set an image away and see what happens... May 26 03:25:42 hrw: some people reported they had to remove the 2.4 card entry from fstab to make the card automount under 2.6. i don't understand how that should happen, but it may be worth a try May 26 03:26:42 ah yes, I did that too May 26 03:27:57 mine doesnt automount since about a week ago, and it worked fine before that May 26 03:28:21 re May 26 03:28:22 thx May 26 03:28:37 XorA: There were changes to udev in recent kernels 2.6.12-rc4 onwards ish May 26 03:31:29 i'd be grateful for a clear policy document May 26 03:31:37 from the kernel, udev, hotplug, pcmcia people May 26 03:31:45 i have no idea how that stuff is supposed to work together now May 26 03:31:56 i had hardly an idea previously, but it's worse now May 26 03:31:56 heh May 26 03:32:07 mickeyl: do they know either ? May 26 03:32:25 ade|desk: yeah, they do. that May 26 03:32:38 makes it worse for us, because they don't have to explain to each other May 26 03:32:52 gits :( May 26 03:43:14 mickeyl: I removed that fstab entry too May 26 03:43:45 mickeyl: also my devices script does not start when udev is in use (like the copy in OE repo) May 26 03:46:53 koen: I tried building a gpe-image, but it still tries to install linux-jlime-sh3-2.6.11-r0 :( May 26 03:47:36 strange May 26 03:47:40 could you file a bug on it? May 26 03:48:25 Arjan: what machine you buildfor? May 26 03:48:45 koen: I could, but I'm using an unsupported machine May 26 03:49:03 Arjan: rm -rf tmp/cache May 26 03:49:07 Arjan: bitbake -i May 26 03:49:12 Arjan: print PREFERRED_PROVIDERS May 26 03:49:16 paste the output May 26 03:49:27 hrw: my machine is not supported (h1940), so I chose target_arch=arm May 26 03:49:35 oooh, bitbake -i May 26 03:49:36 uhm May 26 03:49:49 ~hail mickeyl, our interactive overlord May 26 03:49:51 * ibot bows down to mickeyl, our interactive overlord and chants, "I'M NOT WORTHY!!" May 26 03:49:56 without a PREFERRED_PROVIDER_virtual/kernel you will build a more or less arbitrary kernel May 26 03:49:59 :) May 26 03:50:12 mickeyl: virtual/qte:qte virtual/libqpe:libqpe-opie virtual/libsdl:libsdl-qpe virtual/arm-linux-gcc-initial:gcc-cross-initial virtual/arm-linux-gcc:gcc-cross virtual/arm-linux-g++:gcc-cross May 26 03:50:18 what i suspected May 26 03:50:26 there is no preferred provider for virtual/kernel May 26 03:50:31 hence bitbake has no idea what to build May 26 03:50:37 hence it chooses one arbitrarily May 26 03:50:40 (more or less) May 26 03:50:58 mickeyl: this is more or less the standard configuration May 26 03:51:26 look at our machine configurations May 26 03:51:32 yo May 26 03:51:36 so why does it build a kernel? I already have one and just want the gpe rootfs May 26 03:51:39 see that we set the preferred kernel for every machine there May 26 03:51:39 try ipaq-pxa-2.6 May 26 03:51:40 udev 057 got RUN option for rules.. I installed 058 on husky May 26 03:51:46 ieeks the noka 770 is only omap1 May 26 03:51:48 Arjan: bitbake gpe-_image_ May 26 03:51:55 an image is an image is an image May 26 03:51:56 woglinde: omap 1710 May 26 03:51:57 an image includes a kernel May 26 03:52:02 arm 926 core May 26 03:52:08 Arjan: if you already have a kernel, you need to say so with ASSUME_PROVIDED. May 26 03:52:10 200MHz armv5TE May 26 03:52:12 mickeyl: I did bitbake gpe-image May 26 03:52:21 yes, of course. that's why you get a kernel. May 26 03:52:40 if you think you don't need one, do what pb_ suggests May 26 03:52:53 pb_: my kernel is not from OE, it's outside the tree May 26 03:53:00 indeed May 26 03:53:05 that's what ASSUME_PROVIDED is for. May 26 03:53:22 of course, it is unlikely that your build will actually succeed under those conditions, but if you want to try it then that is the way to do so. May 26 03:53:24 Arjan: you can also do some work and add that kernel into OE May 26 03:53:26 mickeyl: already tried setting assume_provided=virtual/kernel, and that made pcmcia-uitls or something like that fail to compile May 26 03:54:15 hrw: I wish that was possible, I don't have any experience with OE... just started and trying to build gpe May 26 03:54:47 Arjan: look for those kernels in packages/linux/ May 26 03:56:14 hrw: should I just build a random kernel from packages/linux/ and hope OE uses that one to build gpe? May 26 03:56:57 Arjan: experts gave you infos already May 26 03:58:00 I go outside.. its shinny day, I have free day so why I have to be here? May 26 03:58:11 hrw dont know *g* May 26 03:58:13 have fun May 26 03:58:23 outside & sun is evil ;) May 26 03:58:24 no need to hope, just rely on the facts. OE builds stuff against the last kernel that has been built. May 26 03:58:34 I'll be working inside for the rest of the day May 26 03:58:44 so build a kernel that comes close to the kernel on your device and maybe some things work May 26 03:58:48 but I'll have some 1kW lamps to make up for the lack of sun May 26 03:58:52 the external modules won't for sure, however May 26 03:59:16 OE is not supporting patch work builds May 26 03:59:28 We're reconstructing complete things from scratch May 26 03:59:40 that's one of the paradigmes here May 26 03:59:41 (imo) May 26 04:00:10 mickeyl: one last question: if I add my kernel and let defconfig build some modules, will they get into the images automatically? May 26 04:00:23 arjan yes May 26 04:00:32 Arjan: no they won't. please take the time and read _at least one_ machine configurtation May 26 04:00:37 all is written there May 26 04:00:42 oh right May 26 04:00:48 damned May 26 04:00:51 its to hot May 26 04:01:33 woglinde: omap2 would've meant 2007 or so, considering that this device has been 2.5 years in development May 26 04:01:49 mickeyl i know :( May 26 04:02:05 i hate the delay of annoucment and shipping May 26 04:02:05 mickeyl: thanks, I'll do that now. I'm one of those try-first-read-later guys, so for that I'm sorry to ask obvious questions May 26 04:02:30 heh, np May 26 04:05:59 morning May 26 04:06:17 hi reenoo May 26 04:09:04 hey woglinde May 26 04:28:00 hi reenoo_ May 26 04:28:21 hey pb_ May 26 04:31:01 03koen 07 * r1.3401 10openembedded/packages/maemo/hildon-lgpl_0.9.14-2.bb: hildon-lgpl_0.9.14-2.bb: spell LICENSE the correct way May 26 04:31:57 hm is there a vim command to show row ending white spaces? May 26 04:33:16 :set hlsearch May 26 04:33:36 *$ May 26 04:33:38 err May 26 04:33:41 / *$ May 26 04:34:30 renno thanks May 26 04:34:35 ups reeno May 26 04:35:29 Nokia 770 3 hrs battery life? Meh.. May 26 04:35:42 wrobbie hm its okay May 26 04:35:48 fot this device May 26 04:36:40 03mickeyl * r217 10bitbake/lib/bb/shell.py: May 26 04:36:41 moin zecke May 26 04:36:41 - add more commands: edit, clean, rebuild. May 26 04:36:41 - load command history on startup May 26 04:36:41 - save command history on exit May 26 04:36:41 Bug: A bug can be seen when you call rebuild on a package more than once - May 26 04:36:41 somehow the make module no longer calls the tasks then. May 26 04:36:49 heh May 26 04:36:51 seems neat for the rest.. did someone see HW specs somewhere? (going throught site now..) May 26 04:36:59 I read "moin zecke" as the changelog May 26 04:37:04 hehe May 26 04:37:06 nah, not quite :) May 26 04:37:14 hi zecke May 26 04:37:34 wrobbie: omap 1710, unknow other stuff May 26 04:38:54 wroppie maybee here http://www.linuxdevices.com/news/NS5409534614.html May 26 04:38:57 wq May 26 04:38:59 args May 26 04:39:07 bitbake shell? time to read devel ML archives? :) May 26 04:39:19 thanks koen, woglinde May 26 04:41:25 !lart postgres for not ignoring , on the last entry auf creat commandos May 26 04:41:26 * cdbot whips out a sword and chops postgres in half for not ignoring , on the last entry auf creat commandos May 26 04:42:05 wrobbie: i didn't write an announcement for the shell yet. i will do if it's a bit more mature and tested May 26 04:42:29 mickeyl when can we watch tv with bitbake? May 26 04:42:37 :) May 26 04:42:39 woglinde: soon :D May 26 04:42:47 bitbake: the new emacs? May 26 04:43:04 emacs bitbake plugin? May 26 04:43:18 (rewrite in elisp) May 26 04:43:18 a nice thing is that while working on the shell, we see how we need to refactor the bitbake.py and the make.py May 26 04:45:04 mickeyl: re: serial ports. Can't hack it, am simply not into that stuff too well. Will post findings to OE mailinglist. May 26 04:45:21 s/too well/at all/ May 26 04:46:19 it's a kernel thing. it's not mapping the port adresses (it does the irq's tho) May 26 04:46:30 Hertog: ok May 26 04:52:55 <[g2]> mickeyl, pb_ any idea when the last time kergoth did a build world ? May 26 04:53:06 ages May 26 04:53:08 months May 26 04:54:44 <[g2]> I've got my custom Knoppix LiveDVD that has all the OE metadata from a couple days ago and all the tarballs for Openslug/Unslung. I was was thinking about running a build world May 26 04:55:28 <[g2]> It builds Openslug if you give it a swap and file space for "tmp" May 26 04:55:55 <[g2]> any idea how big a "tmp" area I'd need ? May 26 04:56:09 for world build? May 26 04:56:13 <[g2]> nod. May 26 04:56:15 18 gigs or so May 26 04:56:18 including sources May 26 04:56:42 mickeyl: could we create a 'Command' API specification and have a dict that goes from Keyword -> CommandType? May 26 04:56:44 <[g2]> I've got 350MB of sources for Openslug May 26 04:56:54 mickeyl: so one could easily implement new commands? May 26 04:57:05 fetching world will be 7GB or something May 26 04:57:12 or 4GB May 26 04:57:18 * koen should test it May 26 04:58:23 zecke: that'd be nice. we have to do a lot of refactoring for that, i.e. in make.py and bitbake, but it will be worth it in the end May 26 04:59:34 <[g2]> zecke, I'd like to talk to you about the tinderbox thing sometime. With the liveCD/DVDs I'm looking to allow full automated regression testing for OpenSlug May 26 04:59:40 mickeyl: oh, I thought of that limited to the shell but yes doing it globally would make sense as well May 26 04:59:51 03koen 07 * r1.3402 10openembedded/packages/maemo/osso-bttools_0.25.1.bb: Add osso-bttools, part of the maemo.org suite May 26 05:00:23 [g2]: I've too problems to solve, teach tinderbox to know SVN (easy), teach it to be aware of bk May 26 05:00:45 koen: how much is missing? May 26 05:00:54 zecke: about 90% May 26 05:01:07 koen: how much is done? ;) May 26 05:01:14 zecke: what's most urgent at the moment is that we have to continue doing some OO refactorings. bitbake and make.py are wieldly making assumptions about state and order of control flow. this has to be refactored May 26 05:01:24 [g2]: On the client side I will use bitbake with special bbclasses to run the tests May 26 05:01:25 about 25% is in my local repo right now May 26 05:01:54 koen: so it is for real, all source is available May 26 05:02:10 koen: is some library only GPL licensed or is everything LGPL? May 26 05:02:32 koen: planning to release maemo before the nokia device is actually available in shops? :) May 26 05:02:49 as far as I can see it's all lgpl, with some closed source stuff May 26 05:02:58 * [g2] was guessing that would happen :) May 26 05:03:10 Hertog: If I can spare a few more days it will be in OE May 26 05:03:25 eh ;) May 26 05:03:55 * [g2] thinks this was a *great* *bold* move by Nokia May 26 05:04:01 which I can't, but I'm waiting for an answer on my assignment, so I have a bit of time now May 26 05:04:31 koen: Why not wait for the answer on the 'Oude Markt' ;) May 26 05:05:00 koen: all those teensy nice summer dresses ;) May 26 05:05:02 Hertog: nice idea, but I have to work in about an hour :) May 26 05:06:44 zecke: oh and i hate the code duplication in bitbake in showVersion May 26 05:07:35 mickeyl: yupp May 26 05:07:45 mickeyl: we've some more at some other places May 26 05:08:02 mickeyl: but we really should release it 'soon' before breaking it too heavy May 26 05:08:39 zecke: let's fix the -r bug and the make a release May 26 05:08:53 then we can do more structural changes afterwards May 26 05:09:08 i'm not clueful enough to fix the -r though May 26 05:09:16 it breaks too deep inside the core :) May 26 05:09:34 mickeyl: it will be easy and obvious May 26 05:11:26 03koen 07 * r1.3403 10openembedded/packages/maemo/libosso_0.8.9.bb: libosso_0.8.9.bb: ship services and libraries too May 26 05:12:20 what's the "-r bug"? May 26 05:12:49 pb_: parsing a config file before conf bitbake.conf May 26 05:13:05 though I not seen it myself yet May 26 05:13:17 that's what -r is meant to do, iirc. May 26 05:13:19 it's a simple bug that shows a deeper problem May 26 05:13:23 see: May 26 05:13:42 mickey@gandalf:/local/pkg/oe/c7x0$ bitbake -r invalidConfFile nano May 26 05:13:42 Traceback (most recent call last): May 26 05:13:42 File "/local/pkg/oe/bitbake/bin/bitbake", line 789, in ? May 26 05:13:42 cooker.cook( args[1:] ) May 26 05:13:42 File "/local/pkg/oe/bitbake/bin/bitbake", line 640, in cook May 26 05:13:42 self.parseConfigurationFile( os.path.join( "conf", "bitbake.conf" ) ) May 26 05:13:44 File "/local/pkg/oe/bitbake/bin/bitbake", line 599, in parseConfigurationFile May 26 05:13:46 make.cfg = bb.parse.handle( afile, make.cfg ) May 26 05:13:48 File "/local/pkg/oe/bitbake/lib/bb/parse/__init__.py", line 70, in handle May 26 05:13:50 return h['handle'](fn, data, include) May 26 05:13:52 File "/local/pkg/oe/bitbake/lib/bb/parse/ConfHandler.py", line 104, in handle May 26 05:13:54 init(data) May 26 05:13:56 File "/local/pkg/oe/bitbake/lib/bb/parse/ConfHandler.py", line 32, in init May 26 05:13:58 if not bb.data.getVar('TOPDIR', data): May 26 05:14:00 File "/local/pkg/oe/bitbake/lib/bb/data.py", line 127, in getVar May 26 05:14:02 return d.getVar(var,exp) May 26 05:14:04 AttributeError: 'NoneType' object has no attribute 'getVar' May 26 05:14:06 this is an invalid conf file, because it doesn't end with .conf May 26 05:14:08 ok so far May 26 05:14:10 however May 26 05:14:13 it shows some things going wrong with 'd' May 26 05:14:16 because that should never be None in this place May 26 05:14:34 and i suspect there is a good deal of shadowing taking place in these modules May 26 05:14:41 a can of worms May 26 05:15:06 hmm worms May 26 05:15:34 we could simply check if the filename ends with .conf, but that would just mask an inner bug May 26 05:15:54 i couldn't make enough sense of the code paths in data.py and confHandler.py though May 26 05:16:43 having data be the name of a module and sometimes also as per package data struct is very bad May 26 05:16:52 then 'd' is sometimes data and sometimes not May 26 05:16:55 * mickeyl shudders May 26 05:17:27 so, is the problem just that bb.parse.handle() returns None if you feed it something bogus? May 26 05:17:43 if that's the case, bitbake should be detecting that situation and taking evasive action May 26 05:17:47 pb_: sounds like that May 26 05:18:00 and confHandler.handler() is getting called twice in a row once with data being {} first and None second. and that's when it breaks May 26 05:18:03 pb_: but normally it returns the data instance May 26 05:18:06 (but I think bb.parse.handle should really be throwing an exception, not quietly returning some bogus thing) May 26 05:18:18 zecke: right May 26 05:18:30 so, either add a check for None at this point: May 26 05:18:36 pb_: right the exceptions sounds right May 26 05:18:43 for f in make.options.file: May 26 05:18:43 try: May 26 05:18:43 make.cfg = bb.parse.handle(f, make.cfg) May 26 05:18:51 pb_: because otherwise the behaviour would be too asymetric May 26 05:18:53 so that it doesn't clobber the old make.cfg if bb.parse.handle returns None May 26 05:19:04 or, preferably, have bb.parse.handle throw an exception, and catch it in that try block. May 26 05:19:20 or return the 'input' always... May 26 05:19:27 mickeyl: either you decide now or I'll later May 26 05:19:37 yeah, I guess that would work too, but returning the input would make it impossible to detect failure May 26 05:19:45 I think the exception is the best approach May 26 05:19:53 ack May 26 05:20:05 since, generally, if you feed a bogus file to -r, you would expect to get at least a diagnostic. May 26 05:20:31 zecke: i'll leave that to you May 26 05:20:48 i'm currently struggling against a readline completer for bb files :) May 26 05:20:53 (for the shell) May 26 05:20:53 heh May 26 05:21:21 mickeyl: could we enter gödel numbers as input? May 26 05:21:26 http://handhelds.org/scap/port.30416.png May 26 05:21:38 :) May 26 05:21:43 so I could do range(1:) and wait until the right bitbake file was found May 26 05:21:45 cool May 26 05:22:12 I'll check in a fix for the -r thing. I think it just needs a one-line patch in __init__.py and an extra "except" clause in bitbake. May 26 05:22:14 zecke: why not :) May 26 05:22:19 pb_: thanks May 26 05:22:27 pb_: go ahead May 26 05:22:40 reenoo_: what's that? May 26 05:22:52 gtk-webcore May 26 05:22:54 http://www.spiegel.de/netzwelt/politik/0,1518,357512,00.html May 26 05:22:57 reenoo_: sweet! May 26 05:25:19 hm, does bitbake actually trap ParseError at all? May 26 05:26:23 pb_: yupp May 26 05:26:31 it prints conf/bitbake.conf not found May 26 05:26:51 oh, drat, my bitbake checkout is read only May 26 05:26:56 * pb_ looks for a writeable tree May 26 05:27:44 svn+ssh May 26 05:28:16 is there any way to convert an anonymous checkout into a non-anonymous one? May 26 05:28:22 without checking out a whole new tree? May 26 05:29:10 pb_: svn switch May 26 05:29:25 aha May 26 05:30:05 03philb * r218 10bitbake/lib/bb/parse/__init__.py: May 26 05:30:05 have bb.parse.handle() throw ParseError if the input file is not May 26 05:30:05 understood by any handler, rather than quietly returning None May 26 05:30:18 well, that should improve matters a bit. May 26 05:31:31 I still get a bit of python backtrace spew: it seems that bitbake isn't correctly trapping ParseError for -r files. May 26 05:31:45 pb@mebius:~/oe/build-h3900$ bitbake -r bogusFile nano May 26 05:31:45 Traceback (most recent call last): May 26 05:31:45 File "/home/pb/oe/bitbake/bin/bitbake", line 622, in ? May 26 05:31:45 make.cfg = bb.parse.handle(f, make.cfg) May 26 05:31:45 File "/home/pb/oe/bitbake/lib/bb/parse/__init__.py", line 71, in handle May 26 05:31:46 raise ParseError("%s is not a BitBake file" % fn) May 26 05:31:48 bb.parse.ParseError: bogusFile is not a BitBake file May 26 05:32:21 I guess whatever exception trap it applies to the main parse needs to be applied there as well. May 26 05:33:16 yupp May 26 05:50:26 03koen 07 * r1.3404 10openembedded/packages/maemo/ (2 files in 2 dirs): Add osso-gwconnect, part of the maemo.org suite May 26 05:50:53 wow the new bitbake is even faster than the old one May 26 05:52:00 been an update today? May 26 05:52:18 Ken|Jlime: there has been an update May 26 05:52:30 Ken|Jlime: but I just looked at the compile times of our daily builds May 26 05:52:46 Aha May 26 05:52:52 did you get any nice numbers May 26 05:52:53 ? May 26 05:53:41 Ken|Jlime: it saves 20 minutes May 26 05:53:51 out of how long in total? May 26 05:54:14 pb_: 2 hours twenty with the old bitbake May 26 05:54:24 pb_: 2:40-2:60 with the newer one May 26 05:54:37 pb_: 1:50-2:00 with the current one May 26 05:54:47 cool, that's quite an improvement May 26 05:54:55 brb, need to switch off the electricity here for a few minutes May 26 05:55:01 neat May 26 05:55:11 oh getting too hot? May 26 05:57:44 He couldnt handle the exciting new bitbake numbers :) May 26 05:59:44 re May 26 06:00:26 zecke: heh, no, it isn't very hot here. May 26 06:00:32 this is England, remember May 26 06:01:43 pb_: but it is London and even we've 30° May 26 06:01:52 * ade|desk bakes in sunny norfolk May 26 06:02:32 26C here May 26 06:02:46 "warm enough" May 26 06:03:13 zecke: http://www.wunderground.com/global/stations/03571.html May 26 06:03:14 heh May 26 06:04:33 lol May 26 06:25:15 03koen 07 * r1.3405 10openembedded/packages/maemo/ (osso-gwobex_0.26.1.bb osso-gwconnect_0.70.2.bb): May 26 06:25:15 Add osso-gwobex, part of the maemo.org suite May 26 06:25:15 osso-gwconnect_0.70.2.bb: stage some headers May 26 06:25:32 03mickeyl * r219 10bitbake/bin/bitbake: May 26 06:25:32 - remove checking for return if parse-only is set when executing just one BB May 26 06:25:32 - don't sys.exit() for parse-only, but return. This makes dropping into the Python interactive mode possible. May 26 06:26:14 03mickeyl * r220 10bitbake/lib/bb/shell.py: add bbfile completion May 26 06:26:21 woot May 26 06:26:49 try: May 26 06:26:52 bitbake -i May 26 06:26:53 parse May 26 06:26:58 build a May 26 06:26:59 :) May 26 06:27:41 (but see the BUGS entry in shell.py) May 26 06:27:58 mickeyl: pay attention to your dis May 26 06:28:12 mickeyl: I know how exciting it can be May 26 06:28:29 hi May 26 06:31:08 silence May 26 06:31:24 :) May 26 06:31:46 * koen silent in amazement May 26 06:32:16 * mickeyl ponders whether to continue hacking or to get outside in the sun May 26 06:32:34 someone, quick, tie him to his chair! May 26 06:32:43 heh May 26 07:01:27 Hertog: I think his wife reserved such rights... May 26 07:04:00 03koen 07 * r1.3406 10openembedded/packages/maemo/osso-thumbnail_0.3.1-1.bb: Add osso-thumbnail, part of the maemo.org suite. May 26 07:08:43 hmm... nice to see what some reading can do... May 26 07:09:03 I just finished adding h1940 machine and kernel to OE May 26 07:10:56 attach the file/patch to bugzilla and we'll take a look at it May 26 07:11:42 thanks. should I make it a patch or a tgz with the files? May 26 07:13:22 patch for modified files, tgz for new files May 26 07:14:08 * Arjan is happy he could actually do something for one of the projects May 26 07:26:42 03koen 07 * r1.3407 10openembedded/packages/dbus/dbus_0.23.4.bb: Update (old API) dbus to 0.23.4 May 26 07:34:15 03koen 07 * r1.3408 10openembedded/packages/maemo/libosso_0.8.9.bb: libosso_0.8.9.bb: It's pkgconfig with 2 g's May 26 07:57:27 03mickeyl * r221 10bitbake/bin/bitbake: catch parse error in parseConfigurationFile() May 26 08:00:00 03mickeyl * r222 10bitbake/lib/bb/shell.py: some fixes May 26 08:00:21 zecke_: if you don't have any other things to tweak, i'd say feel free to tag and release this as bitbake 1.3.0 May 26 08:00:34 03CoreDump 07 * r1.3410 10openembedded/packages/altboot/ (altboot_cvs.bb altboot_20050527.bb): Complete Merge May 26 08:00:38 03CoreDump 07 * r1.3409 10openembedded/packages/altboot/altboot_cvs.bb: Merge May 26 08:00:41 03CoreDump 07 * r1.3395.1.1 10openembedded/packages/altboot/ (altboot_cvs.bb altboot_20050527.bb): altboot: Add documentation May 26 08:00:54 mickeyl: awesome May 26 08:00:55 i have some more invasive changes in the pipeline which should be done later May 26 08:01:32 and that is enough computing for today :) May 26 08:01:39 l8er May 26 08:04:13 03koen 07 * r1.3408.1.1 10openembedded/packages/maemo/ (3 files): osso-thumbnail_0.3.1-1.bb, hildon-lgpl_0.9.14-2.bb, libosso_0.8.9.bb: stage headers May 26 08:04:50 ypu May 26 08:05:14 is there a reason for the fact we don't auto-stage -dev packages? May 26 08:05:19 you're pretty busy with this maemo stuff :) May 26 08:05:47 I can't let scratchbox take all the attention :) May 26 08:05:58 hehe May 26 08:09:11 mickey_away: cya May 26 08:09:22 mickey_away: ok deep feature freeze for bitbake now May 26 08:09:30 deap even.. May 26 08:09:49 ~lart Qt's clipboard May 26 08:09:49 * ibot pours hot grits down the front of Qt's clipboard's pants May 26 08:10:39 koen: my files are on bugzilla, id 40 May 26 08:10:51 let's hope someone has time to look at them soon :) May 26 08:10:51 okidoki May 26 08:10:58 going home now; bye all May 26 08:11:22 Arjan: cya May 26 09:34:52 does anyone here remember the option for /etc/ipkg.conf to move /usr/lib/ipkg/lists to another location? May 26 09:36:06 lists_dir ext /path/to/lists May 26 09:37:16 koen: thx! May 26 09:57:00 currently portmap is beeing started both in rcS.d and rc[2-5].d. Any thoughts which should be removed? Is it important to launch portmal in rcS.d? May 26 09:57:26 do we even need portmap? May 26 09:57:49 well, I dunno. It's in the default image though May 26 09:58:07 i tend to remove the rc[2-5].d startup links May 26 09:58:12 CoreDump|home: if part of your filesystem is nfs mounted, then i could see it being necessary to start portmap early to ensure that it succeeds May 26 09:58:13 there was something about portmap a while back May 26 09:58:31 kergoth: agreed May 26 09:58:54 * CoreDump|home remove the other links May 26 10:07:11 koen: ping May 26 10:07:15 pong May 26 10:07:33 the rosetta thing in gpe, is that like a second generation xstroke or something different? May 26 10:07:44 something different May 26 10:07:51 ah May 26 10:07:54 xstroke uses an own alfabet May 26 10:08:09 rosetta is pure handwriting recognitions May 26 10:08:12 oh? May 26 10:08:22 like wince block recognizer May 26 10:08:27 sweet May 26 10:08:34 work well? May 26 10:08:58 it did ok when I tried it May 26 10:09:18 Any ideas what the most popular input method in gpe is? May 26 10:09:33 apart from thge hardware keyboard May 26 10:09:40 no idea May 26 10:09:46 i'm a keyboard person May 26 10:09:58 back when I used X, xstroke was all the rage, but it's been a while :p May 26 10:10:19 treke: http://projects.o-hand.com/matchbox/ May 26 10:10:24 top news posting May 26 10:10:39 mallum has rewritten xstroke May 26 10:10:42 hehe May 26 10:12:04 * koen tries to add 94 patches to the maemo gtk May 26 10:13:55 poor koen May 26 10:14:41 awk +tr will be my friend May 26 10:21:21 treke: ls | awk '{print "file://" $1 ";patch=1"}' May 26 10:21:34 now lets see how good quilt really is May 26 10:27:02 koen: heh May 26 10:27:22 morning treke May 26 10:28:30 reenoo_: ls | awk '{print "\tfile://" $1 ";patch=1 \\"}' is a bit better May 26 10:28:45 pb. May 26 10:30:02 ~lart kde people for changing the List-Id without further notice... May 26 10:30:02 * ibot installs PocketPC on kde people's PDA for changing the List-Id without further notice... May 26 10:31:12 I think you got the further notice, its the prior notice you were missing... :) May 26 10:32:10 heh May 26 10:33:30 CosmicPenguin: without any *readable* notice May 26 10:40:06 ~lart "wrong kernel" message May 26 10:40:06 * ibot burns "wrong kernel" message to a crisp with a laser May 26 10:46:48 NOTE: package gtk+-2.6.4-1.osso7-r1: task do_patch: completed May 26 10:46:59 koen@dominion:/data/build/oe/openembedded/packages/gtk+$ grep patch gtk+_2.6.4-1.osso7.bb | wc -l May 26 10:47:02 104 May 26 10:47:05 03CoreDump 07 * r1.3413 10openembedded/packages/meta/opie-image-ext2loop.bb: Added opie-image-ext2loop.bb: Identical to opie-image but creates an ext2 image file to be used w/ altboot or similar scripts May 26 10:47:07 I think I've set a record :) May 26 10:47:08 03CoreDump 07 * r1.3412 10openembedded/packages/ (5 files in 5 dirs): (log message trimmed) May 26 10:47:08 initscripts_1.0.bb: May 26 10:47:08 initscripts: Bump PR May 26 10:47:08 sharp-sdmmc-support_2.4.18-rmk7-pxa3-embedix.bb: May 26 10:47:08 sharp-sd: Bump PR May 26 10:47:10 portmap_5-9.bb: May 26 10:47:11 portmap: Bump PR May 26 10:47:33 Good evening May 26 10:48:01 hey SirFred May 26 10:55:23 CoreDump|home: uh? is doing "IMAGE_FSTYPES += ext2" too difficult? i dont see the point in a seperate .bb for it May 26 10:57:08 hmm? my local.conf is always set to jffs2, like most Z devs I'd guess. With a seperate .bb file you don't have to edit opie-image.bb to build an ext2 image once in a while May 26 10:57:18 edit opie-image.bb? May 26 10:57:21 what are you talking about? May 26 10:57:37 IMAGE_FSTYPES = "jffs2 ext2" May 26 10:57:40 ta da, done May 26 10:57:48 you dont have to touch any .bb fileas May 26 10:57:50 files May 26 10:58:25 I do, since I mostly use bitbake -b ;) May 26 10:58:48 eh? bitbake -b still reads local.conf. May 26 11:00:11 ok, it does. On the next compile you'll need to wait ages for the parsing since you've touched (needlessy) local.conf. But ok, I'll trash the bb. May 26 11:01:16 needless my ass. pointlessly hardcoding the filesystem type into a .bb is both needless and pointless. May 26 11:01:24 and that was redundant, but you get the point May 26 11:01:34 heh May 26 11:01:59 yeah. setting stuff like IMAGE_ROOTFS_SIZE_ext2 in a .bb is in fairly poor taste, as well. May 26 11:02:11 IMO it isn't. It _is_ easier / less work that way to build such an image once in a while. May 26 11:03:16 for you, possibly, and you are welcome to keep that .bb file on your local disk. May 26 11:03:25 and since the bb is even calles "ext2loop" I can't detect any poor taste with hard-coding it to be ext2 May 26 11:03:29 you are even welcome to use it in the privacy of your own home. May 26 11:03:47 don't try to build a h3900 ext image, some genius made -p default for all image types May 26 11:04:02 03CoreDump 07 * r1.3414 10openembedded/BitKeeper/deleted/.del-opie-image-ext2loop.bb~1275262c1f290f89: May 26 11:04:02 .del-opie-image-ext2loop.bb~1275262c1f290f89: May 26 11:04:02 Delete: packages/meta/opie-image-ext2loop.bb May 26 11:04:27 koen: pardon? May 26 11:04:33 pb_: oh you can count on that May 26 11:05:03 pb_: something like IMG_COMMAND_append = "-p " May 26 11:05:14 instead of IMG_COMMAND_jffs2_append May 26 11:05:24 in what file? May 26 11:05:35 probably h3900, otherwise ipaq May 26 11:05:57 oh yeah May 26 11:05:59 EXTRA_IMAGECMD_h3900 = "-e 0x40000 -p" May 26 11:06:07 -p is the same as --pad, right? May 26 11:06:14 no idea May 26 11:06:36 it worked till ext2tools implemented stricter checking May 26 11:06:47 and failed on the -p May 26 11:06:59 I think that line should be EXTRA_IMAGECMD_jffs2 += "-e 0x40000". May 26 11:07:17 there's no point having the MACHINE override in there, since h3900.conf is not even parsed on any other MACHINE. May 26 11:07:33 and, assuming -p does mean --pad, that's in the default command that bitbake.conf supplies. May 26 11:07:34 ah yes :) May 26 11:24:45 NOTE: package gtk+-2.6.4-1.osso7: completed May 26 11:30:08 * koen whistles innocently May 26 11:30:13 03koen 07 * r1.3415 10openembedded/packages/gtk+/ (107 files in 2 dirs): Add gtk+-2.6.4-1.osso7 + one hundred and four (104) patches, part of the maemo.org suite. (I believe this sets a record, try and beat that, RP ;)) May 26 11:36:04 03reenoo 07 * r1.3400.1.1 10openembedded/packages/linux/handhelds-sa_2.4.19-rmk6-pxa1-hh39.bb: handhelds-sa_2.4.19-rmk6-pxa1-hh39: DEFAULT_PREFERENCE = "-1" for now May 26 11:38:13 Nice May 26 11:38:18 koen: see - quilt does rock after all May 26 11:38:26 heh May 26 11:38:42 awk and vi were all the tools I used May 26 11:38:52 and head, tail and cat May 26 11:39:21 bah May 26 11:39:47 they had one line of osso weirdness left, so I edited the patch May 26 11:43:56 koen: That's cheating - you didn't write all the patches :) May 26 11:44:02 :) May 26 11:45:33 RP: do you have any idea what's wrong with the serial ports in >2.6.11 on z7x0? (or... do you even know something goes wrong there?) May 26 11:46:55 03koen 07 * r1.3415.1.1 10openembedded/packages/gtk+/gtk+_2.6.4-1.osso7.bb: gtk+_2.6.4-1.osso7.bb: fix MAINTAINER May 26 11:50:51 koen: with +100 patches one could gtk+ into qt ;) May 26 11:51:27 zecke: one sure could :) May 26 11:58:42 mickey_away: 1.3.0 really? May 26 11:59:04 kergoth: we would like to release a new version of bitbake 'soon' May 26 12:00:36 03zecke123 * r223 10bitbake/ (ChangeLog lib/bb/__init__.py): May 26 12:00:36 bitbake: May 26 12:00:36 -Prepare the 1.3.0 release of bitbake May 26 12:02:23 Hertog: Have you looked at mickeyl May 26 12:02:28 mickeyl's patch May 26 12:02:53 zecke: Yes, go for 1.3.0 :) May 26 12:03:30 RP: eh, looked at it, but as I said to mickeyl, I'm out of my depth there, am not much of a coder, much less a kernel hacker ;) I do have some 'evidence' tho. May 26 12:04:26 Hertog: What's the evidence? May 26 12:05:28 RP: nothing much.. Kernel detects the serial ports fine. setserial shows the right irq's are assigned, but it is impossible to set the ioports to the right adresses May 26 12:05:53 Hertog: Which kernel is this exactly? May 26 12:06:06 and which serial ports for that matter? May 26 12:07:14 Linux version 2.6.12-rc4-mm2- (but the 'official' 3.5.3 shows the same behaviour) and it's for ttyS[0,1,2], or as the kernel tells me, a FFUART, a BTUART and a STUART May 26 12:08:37 and how are you trying to use the ports? May 26 12:09:07 There is at least some functionality there as the serial consoles work for me May 26 12:09:45 It could be you're trying to access IO lines not present on the port. It could also be that the serial console (which is enabled by default) is interfering with your serial device if you're using FFUART May 26 12:10:54 I'm not trying to use them, as such, but problems occur when I insert a gps or bt card, those cards get ttyS3 (in dmesg and on fbconsole) , but as cardmon thinks ttyS0-2 are unassigned it tries to do magic on those instead of on ttyS3 May 26 12:11:27 RP: http://pastebin.ca/12731 May 26 12:12:13 That sounds rather like cardmon (whatever that is) is broken. May 26 12:12:53 Hertog: Ok, you need to work out why cardmon thinks ttyS0-2 are unassigned May 26 12:13:00 pb_: could be. As said, I'm out of my depths here, but since setserial tells that the UARTS are unknown... May 26 12:13:07 pb_: and unsettable May 26 12:13:26 setserial just doesn't understand pxa serial ports. They are unsettable because there are no configurable parameters. May 26 12:13:44 Hertog: unknown is not the same as unassigned May 26 12:13:44 You can't change the base address or the irq: those are fixed by the design of the chip. May 26 12:14:13 RP: that sounds logical ;) May 26 12:14:25 03koen 07 * r1.3418 10openembedded/packages/maemo/ (3 files): osso-thumbnail_0.3.1-1.bb, osso-gwobex_0.26.1.bb, hildon-lgpl_0.9.14-2.bb: DEPENDS: gtk+ -> gtk+-2.6.4-1.osso7 May 26 12:15:03 So it is indeed cardmon whichis bonkers. OK, will go on with my quest in that direction ;) May 26 12:15:13 Hertog: You need to find out why cardmon chooses ttys0-2. If that's due to bad information given out by the kernel, we can fix that. Its more likely cardmon makes assumptions it shouldn't May 26 12:15:39 it broken on ipaq/2.4 too May 26 12:16:33 RP: somewhere I am glad it is cardmon... that code I can at least understand a bit May 26 12:16:40 whereas the kernelstuff... hats of RP :) May 26 12:17:06 Hertog: Its just a question of what you're use to ;-) May 26 12:17:45 ok, off to add some printflines to cardmon then :) May 26 12:19:52 I got ticked off by setserial telling me that the port was 0x0000 instead of 0x401 (and the irq being correct) ah well... one learns May 26 12:20:26 Where do you get 0x401 from? May 26 12:21:15 The port is actually at 0x40100000, but setserial only prints the low 16 bits. May 26 12:22:15 mostly because the base address is only interesting for ISA ports. May 26 12:22:22 dmesg tells me it's 0x40100000, just abrevated it May 26 12:22:51 * Hertog should not forget to add the significant zero's May 26 12:23:06 heh May 26 12:25:05 hmm, while searching I have seen some patches to setserial floating around so that it knows about those type of uarts, are those intresting for OE? (doubt it) May 26 12:25:22 interesting even May 26 12:25:38 If they're good patches, I think it would be interesting May 26 12:25:56 If they make an OE supported machine work ,they're intresting - that the whole point of a distro after all May 26 12:26:44 CosmicPenguin: I guess it's not about making a machine 'work' as such, more a kind of 'reporting' stuff correctly May 26 12:27:12 same difference May 26 12:32:21 'morning' all May 26 12:32:29 hi hrw May 26 12:32:36 03koen 07 * r1.3419 10openembedded/packages/maemo/hildon-lgpl_0.9.14-2.bb: hildon-lgpl_0.9.14-2.bb: stage some libs too May 26 12:33:13 hey hrw May 26 12:34:12 http://www.imdb.com/title/tt0352418/ - weird movie May 26 12:39:43 I'll bet May 26 12:39:59 CosmicPenguin: but nice May 26 12:40:47 I'll take your word for it May 26 12:40:58 when I read people^wzaurus 'users' on OESF forums I want to kill some of them for lack of brain... May 26 12:41:36 hrw: like I said, your IQ drops 30 points when you buy sharp hw May 26 12:42:08 koen: sometimes it looks like 30 is what left May 26 12:42:08 koen: ... May 26 12:42:16 whow, Am I the first to have a negative IQ then? May 26 12:43:23 heh.. tomorrow I have to go to work ;( May 26 12:49:24 Ciao all May 26 12:50:04 reenoo_: Hi. Thanks for the mail in Gtk+WebCore mailinglist. I'm the "maintainer" of it May 26 12:56:08 hi Pigi May 26 12:56:19 hi hrw May 26 12:56:27 brb May 26 12:57:58 can someone write euro sign? May 26 12:58:03 ? May 26 12:58:10 ? May 26 12:58:14 like that ? May 26 12:58:16 ok. thx May 26 12:58:19 np May 26 12:58:26 heh May 26 12:58:26 you're welcome :) May 26 12:58:27 still no luck with it here May 26 12:59:33 that shouldn't be a big problem. Better to be far from euro's :) May 26 13:00:27 Pigi: I'm switching my debian machines to pl_PL.UTF8 May 26 13:00:58 eheheh I suspected something like that, but anyway, if you can, keep away from euro's :) May 26 13:02:25 sage advice May 26 13:05:09 Pigi: before 2008 (iirc) I will still use pln May 26 13:05:46 so you're safe for at least other 2,5 years :) May 26 13:05:59 heh May 26 13:06:25 also the guys in uk are yet safe ;) Hi pb_ May 26 13:06:27 hmm. it seems that I have to reboot to get system working in utf8... May 26 13:06:48 hi pigi :-) May 26 13:07:11 yes, it seems there is no prospect of the UK signing up to the euro any time soon May 26 13:10:55 I think the English delight in the uniquness of their monetary system May 26 13:11:29 CosmicPenguin: if that where so, they wouldn't have gone metric with it :) May 26 13:11:50 hmm.. someone tried to do fresh cvs checkout with current bitbake? May 26 13:13:41 bb.mkdirhier()?? May 26 13:14:41 ok. found May 26 13:18:19 looking at last bugs reported by develusers I vote for adding bitbake version to infos printed on start of build May 26 13:21:12 pb_ you know why my patches about kernel compile in oe haven't been applied ? May 26 13:21:16 I was fetching from cvs something which was added 2 days later... May 26 13:21:43 hrw: schurig named that function... May 26 13:22:14 <[g2]> hey kergoth I was thinking about kicking off a world build from my Custom build Knoppix DVD that's got all the OE metatdata, bitbake, etc... and all the tarballs for Openslug/Unslung May 26 13:22:41 [g2]: how do you want to test the built is okay? May 26 13:23:02 Pigi: sorry, no, I don't know May 26 13:23:09 ok, np May 26 13:23:10 it might just be that nobody has looked at them yet. May 26 13:23:12 zecke: name is ok for me - I had other bug May 26 13:23:28 hrw: but you solved it? May 26 13:23:34 (sorry I'm lagging) May 26 13:23:40 I'm deep into perl *shouder* May 26 13:23:58 poor guy May 26 13:24:11 zecke: solved - upped CVSDATE May 26 13:24:47 kergoth: I'm using SVN::Client to have a SVN Tinderbox VC_Display :} May 26 13:24:49 mickey_away: opie-pcmciaapplet try to install pics/pcmcia/*.png which does not exist May 26 13:25:01 <[g2]> zecke, good question. Eventually, I think there can be custom regression tests by project May 26 13:25:14 <[g2]> initially, just compiling it all is a feat May 26 13:26:17 zecke: cute May 26 13:26:24 mickey_away: it use cardmon/pcmcia icon... May 26 13:26:29 it's a shame tinderbox is all written in perl. May 26 13:27:25 pb_: yes, lack of documentation, lack of examples, lack of common formating... May 26 13:27:30 but yes perl is the biggest burden May 26 13:27:33 But it was my impression that you could still do a client in any scripting you want - assuming you can do the e-mailing bits May 26 13:27:46 CosmicPenguin: right May 26 13:27:55 CosmicPenguin: I'll use bitbake for the client May 26 13:28:19 bitbakes executes its' own regression suite May 26 13:29:29 CosmicPenguin: sure, the client is no problem. but all the server-side stuff (which is where most of the complexity is) is a mass of crufty and obscure perl. May 26 13:29:40 zecke: which of the tinderbox variants are you using? May 26 13:30:33 pb_: tinderbox2 right from mozilla.org May 26 13:30:37 pb_: I'll agree with all of that May 26 13:31:06 pb_: tinderbox3 seems unfinished and I couldn't find sourcecode for the tinderbox3 of osdl May 26 13:32:04 grr... why I can't I seem to get bitbake to accept svn as a preferred version? May 26 13:34:56 zecke: yah, that was my experience as well May 26 13:35:27 jg once told me that tinderbox2 was an evolutionary dead end, and tinderbox3 was where all the action was at, but as far as I could tell tinderbox3 was also abandoned. May 26 13:36:19 CosmicPenguin: the PREFERRED_VERSION needs to match PV, which is probably "0.0cvsXXXXX" or something. Setting it to "svn" won't have any useful effect. May 26 13:43:35 03CoreDump 07 * r1.3414.1.1 10openembedded/packages/busybox/busybox_1.00.bb: Unb0rk busybox .bb May 26 13:44:27 03CoreDump 07 * r1.3421 10openembedded/packages/busybox/busybox_1.00.bb: busybox: bump PR May 26 13:45:49 pb_: crap May 26 13:47:30 pb_: tinderbox3 seams dead as well, the only thing that speaks for tinderbox is, it is actually used May 26 13:47:37 man, this maemo stuff is tedious May 26 13:47:38 and it seams to scale May 26 13:47:41 after an "svn up" I'm getting the following error: May 26 13:47:43 +mhentges@oe-head:~/OpenEmbedded/bitbake >bitbake --version May 26 13:47:44 File "/home/mhentges/OpenEmbedded/bitbake/bin/bitbake", line 647 May 26 13:47:44 <<<<<<< .mine May 26 13:47:44 ^ May 26 13:47:44 SyntaxError: invalid syntax May 26 13:47:51 any clues? May 26 13:47:55 CosmicPenguin: conflict May 26 13:48:01 CoreDump|home: looks like an unmerged conflict. May 26 13:48:04 heh May 26 13:48:10 hmm May 26 13:48:27 CosmicPenguin: either deconflict it, or just rm the file and svn up May 26 13:48:42 CosmicPenguin: sorry May 26 13:48:49 svn revert file May 26 13:49:02 CoreDump|home: svn resolved palestinian-conflict May 26 13:49:52 03koen 07 * r1.3420 10openembedded/packages/maemo/hildon-lgpl_0.9.14-2.bb: hildon-lgpl_0.9.14-2.bb: fix install and stage May 26 13:50:57 ahh, svn revert + svn up seems to have resolved the problem. Thanks! May 26 13:53:23 koen: you should probably consider using oe_libinstall and autotools_stage_includes in your do_stage methods. May 26 13:53:56 not that it really matters, but it would be good form May 26 13:54:49 CoreDump|home: eh, that changeset that you checked in to busybox looks kind of bogus. May 26 13:54:59 -pkg_prerm_${PN} () { May 26 13:54:59 +pkg_prerm () { May 26 13:55:00 i know.. May 26 13:55:07 this will cause the mount stuff to appear in all the subpackage postinsts. May 26 13:55:14 bitbake choked in the _${PN} May 26 13:55:20 your bitbake is too old May 26 13:55:21 dunno why May 26 13:55:33 you need a newer version. May 26 13:55:41 yeah, just did an update May 26 13:56:12 anyways, since it's the same thing.. ;) May 26 13:56:23 it's not May 26 13:56:29 hmm? May 26 13:56:32 back that cset out May 26 13:56:48 CoreDump|home: what do you mean by "it's the same thing"? May 26 13:57:09 pkg_prerm and pkg_prerm_${PN} are not equivalent, if that's what you were saying. May 26 13:57:12 something() == something_${PN} or not? May 26 13:57:14 no May 26 13:57:19 how so? May 26 13:57:36 pkg_prerm() will apply globally to everything in ${PACKAGES}. May 26 13:57:43 ah crap May 26 13:57:45 pkg_prerm_FOO() will only apply to that one package. May 26 13:58:36 so, with your change, busybox-httpd and busybox-udhcpd will both end up with the /bin/mount update-alternatives stuff in their postinsts, which is not what we want. May 26 13:58:38 PACKAGES =+ "${PN}-httpd ${PN}-udhcpd", sorry didn't see that before May 26 13:58:53 pb_: oelibinstall created foo.so/ (as a dir) May 26 13:58:58 but I'll look at it May 26 13:59:40 koen: that probably means you invoked it wrong. You need to give it the directory to install into, not the pathname of the destination library, since it installs multiple files. May 26 13:59:56 aha May 26 14:00:07 will test that after my toolchain gets rebuilt May 26 14:00:14 there are numerous examples of (hopefully correct) oe_libinstall usage in other packages. May 26 14:00:26 I imagine a few seconds with grep will turn some up for you to look at. May 26 14:00:42 is there a compelling reason not to auto stage includes and libs? May 26 14:01:34 no; just historical inertia, really. May 26 14:02:06 I think it would be reasonable to put autotools_stage_includes() in a do_stage() method for autotools.bbclass, and maybe even base.bbclass. May 26 14:02:13 I'm also curious why the .so only ends up in .libs/ May 26 14:02:19 likewise some similar function for libraries. May 26 14:02:36 koen: that's just the way libtool works May 26 14:03:01 yeah but some .so's end up in src/ others don't May 26 14:03:31 afaik, libtool will always put the .so files in .libs/. May 26 14:03:48 03CoreDump 07 * r1.3419.1.3 10openembedded/packages/busybox/busybox_1.00.bb: busybox: Undo previous cset, problem was caused by a too old bitbake version May 26 14:03:51 if it's landing in the parent directory, either the makefile must be doing something special, or it isn't using libtool. May 26 14:04:00 ah, I see May 26 14:04:27 actually, it might be that libtool puts the .so files in src/ if you've told it not to build a static library. May 26 14:04:45 the ostensible reason for the whole .libs thing seems to be to segregate the stuff relating to static and shared libs. May 26 14:06:23 this kind of thing is the main argument in favour of writing an autotools_stage_libs() kind of thing, since "make install" is the only thing that is guaranteed to know how to fish the libraries out of whereever libtool has put them. May 26 14:08:01 CoreDump|home: thanks for reverting that change. May 26 14:08:37 np, it was bogus after all May 26 14:09:10 pb_: implementing the autostage thingie wouldn't break stuff, right? May 26 14:09:22 http://bugs.busybox.net/view.php?id=23 - can we have it in busybox? May 26 14:09:24 do_stage in the .bb would still override it? May 26 14:09:29 0000023: patch for ^H in vi's insert mode May 26 14:11:08 its annoying when I press Backspace in busybox/vi and have to go ESC, then x to remove ^H May 26 14:11:17 nite all May 26 14:11:39 is it really a bug in busy box or just the terminal definition? May 26 14:12:41 koen: right May 26 14:12:43 treke: good question.. May 26 14:13:43 treke: iirc, busybox laughs in the face of terminfo; it just has its own set of expectations about how terminals will behave. May 26 14:13:55 the following patch fixes vi so that ^H and DEL will both properly backspace when you're in insert mode, whether or not they're your tty erasechar. this matches the behavior of vile and vim. (the patch also moves some whitespace and fixes a comment typo.) May 26 14:14:01 pb_: fun May 26 14:14:07 So there you ahve it May 26 14:14:15 so itsounds like a busy box bug :p May 26 14:14:48 treke: it does make a certain amount of sense, in that 95% of terminfo exists only to support terminals that have been extinct for three decades. May 26 14:14:59 hehe May 26 14:14:59 bye May 26 14:15:24 except on one persons desk May 26 14:15:34 who will promptly scream very loud if it stops working May 26 14:15:43 I would like to have HP2623A terminal - my first way to use internet..... May 26 14:15:46 I suspect in a lot of cases, there are literally no users. May 26 14:15:56 cu May 26 14:16:28 the terminfo people seem to feel that part of their remit is to act as a historical archive of old terminal systems, which is laudable enough in itself but not really what you want on a handheld system. May 26 14:17:13 bye hrw|gone May 26 14:17:36 hrw|gone: is that an envizex? May 26 14:18:07 nowadays, I think it's a perfectly defensible position to assume that everyone will be using a terminal that understands vt100 codes, at a minimum. May 26 14:21:57 any good reason to have all four of the possible io schedulers in the kernel? May 26 14:22:26 (don't know how much space they take?) May 26 14:22:29 no, that does sound a bit excessive May 26 14:22:52 wow my VC_svn.pm does something May 26 14:23:00 cool May 26 14:23:02 zecke: congrats May 26 14:23:05 lol Out of memory! May 26 14:23:10 so some infinite loop May 26 14:23:18 zecke: if I were you, I'd not bother writing a bk front end May 26 14:23:20 Hertog: I suspect that just the deadline scheduler would be sufficient. May 26 14:23:49 should I file bugreport? May 26 14:24:57 pb_: could you have a look at the autostage business? May 26 14:25:29 Hertog: good idea May 26 14:25:49 'k will do so in the morning, to bed now. G'night May 26 14:26:26 koen: maybe you could check with kergoth and mickey first. Those guys would need to buy into a change like that. May 26 14:27:42 I'll mail the list May 26 14:28:31 good plan May 26 14:31:28 * koen clicks send May 26 14:31:47 grrr.... May 26 14:31:50 damn svn and proxies May 26 14:31:54 lart them right to hell May 26 14:37:11 koen: how do you find out stageable items? (reuse package_split logic?) May 26 14:37:46 check for .so files is ${D}${libdir} May 26 14:37:52 that owuld be my guess May 26 14:38:59 I think kergoth, mickeyl and pb_ will have some more ideas how to do that May 26 14:39:27 yeah, you'd stage ${libdir}/*.so*, ${libdir}/*.a, and maybe the equivalent for /lib. May 26 14:39:45 autotools_stage_includes() just grabs everything from ${includedir}. May 26 14:40:06 obviously, packages with more complex staging requirements will be free to write do_stage() by hand, just like they always have been May 26 14:41:45 pb_: currently I've no objections May 26 14:42:05 zecke: very good. do keep me posted if that situation changes for the worse. May 26 14:42:06 koen: actually I've asked myself the same question a couple of times already May 26 14:43:27 and, arguably, if we're going to auto-stage headers and libraries, we might as well make "inherit pkgconfig" the default as well. May 26 14:43:47 yes May 26 14:43:56 that would prevent me from misspelling it May 26 14:44:02 ~lart pkconfig May 26 14:44:03 * ibot hooks into a hydrant and hoses pkconfig down May 26 14:44:17 hah May 26 14:44:52 bitbake continues happily with that typo May 26 14:44:53 yeah, there's a separate discussion about whether bitbake should diagnose "inherit bogusclass". May 26 14:45:12 it deliberately doesn't diagnose "include bogusfile", but I think you can make a good argument for inherit being different. May 26 14:46:41 again, that'd be a policy change that requires buy-in from the uberhackers. May 26 14:55:48 koen@dominion:/data/build/oe/openembedded/packages/gnome$ bitbake -b siobhan_0.6.bb May 26 14:55:48 NOTE: exceptions.AttributeError:'NoneType' object has no attribute 'group' while evaluating: May 26 14:55:51 http://ftp.gnome.org/pub/GNOME/sources/siobhan/${@gnome_verdir("0.6")}/siobhan-0.6.tar.bz2 May 26 14:56:00 gooo May 26 14:56:08 * koen guesses gnome.bbclass broke with latest bb May 26 14:56:37 no, gnome.bbclass doesn't understand version strings with only two components. May 26 14:57:46 it's expecting to see PV="0.6.0" or something, in which case it would look for /pub/GNOME/sources/siobhan/siobhan-0.6/siobhan-0.6.0.tar.bz2 May 26 14:58:29 if your directory layout doesn't match that, you'll have to set the SRC_URI by hand. May 26 14:58:52 fwiw, i'm all for bb not ignoring classes it can't pick up May 26 14:59:00 i'd welcome a switch for that though May 26 14:59:08 makes sense May 26 14:59:30 * mickeyl going soapbox May 26 14:59:31 http://www.oesf.org/forums/index.php?act=ST&f=69&t=12864&st=0#entry81683 May 26 15:00:23 heh May 26 15:00:33 my svn stuff gets further May 26 15:01:07 cool. what exactly are you doing with svn ? May 26 15:01:41 mickeyl: get all the commits done, and sort them by date May 26 15:01:53 mickeyl: so if something breaks you can jump to the commit and see who broke it May 26 15:02:01 more or less like bonsai does with cvs, I guess May 26 15:02:02 ah that'd be handy May 26 15:02:13 cvsblame :) May 26 15:02:18 right May 26 15:02:33 It should automatically connect to #oe, lart the person and then leave May 26 15:02:59 hehe May 26 15:03:53 mickeyl: http://tinderbox.mozilla.org/showbuilds.cgi?tree=Firefox May 26 15:04:18 hmm cool May 26 15:04:33 mickeyl: yeah bonsai for poor May 26 15:05:11 the MozillaTest tree is usually a more interestng one to look at, since that build is usually busted. May 26 15:05:18 :) May 26 15:05:25 Firefox is almost always green, which is a bit dull :-} May 26 15:05:54 indeed. i can very well remember opie being red in the early days of your tinderbox setup May 26 15:05:59 heh, right May 26 15:06:21 mickeyl: all images were read May 26 15:06:39 of course they were, but pb_ hacked tinderbox May 26 15:06:58 if (gpe == red) gpe = green; May 26 15:07:10 :D May 26 15:08:08 ah, my hack wasn't that sophisticated. I just had a static assignment: gpe=green, opie=red. May 26 15:08:28 heh May 26 15:08:54 cvsblame was easy too, I just pointed that to a static page that said "br1". May 26 15:08:59 haha May 26 15:09:01 no kidding May 26 15:10:27 anyone actually looked at the 100 patches to gtk? May 26 15:10:36 heh, not me May 26 15:10:37 what kind of patches are that? features or really bugfixes ? May 26 15:10:53 stuff needed to build the rest of meamo May 26 15:11:06 hmm so features then. which makes it not quite a drop-in replacement for gtk right? May 26 15:11:12 is is an older version of gtk, so it won't get pickup by default May 26 15:11:20 I think they're a mixture of features, bugfixes and policy changes. May 26 15:11:27 mickeyl: both May 26 15:11:40 hmm. that leaves a grain of salt to me May 26 15:11:47 it's hard to read the patches, though, since they seem to be split by file rather than by functional set. May 26 15:11:49 sounds like a permanent fork then May 26 15:11:56 mickeyl: adoptions for a PDA, smaller widgets May 26 15:12:04 they want to push upstream May 26 15:12:09 mickeyl: input method changes, disabling tooltips, accells... May 26 15:12:10 i hope they do May 26 15:12:23 since our gpe patches overlap :) May 26 15:12:27 mickeyl: also new features and some look like bugfixes May 26 15:16:39 wow my VC display works May 26 15:16:56 ~praise zecke May 26 15:16:57 All hail zecke! May 26 15:17:02 very good May 26 15:19:16 :) May 26 15:19:25 how do I catch exceptions in perl? May 26 15:21:36 ooops May 26 15:21:40 runaway uclibc config May 26 15:22:19 only 12GB for the log file May 26 15:22:21 I still remember the 90GB logfiles when Kconfig ran wild in the Old Days May 26 15:22:46 it go boom May 26 15:23:03 yeah May 26 15:23:30 I had to keep my fingers crossed when I fired up a build when I was on holiday May 26 15:24:07 when package count was below 1000 May 26 15:24:20 hah May 26 15:24:23 those good ole days May 26 15:26:06 when I didn't go to strange foreign places to drink with strange foreign hackers May 26 15:26:37 my buildbox just crashed after 150 day uptime :\ May 26 15:27:18 'night all May 26 15:28:22 Got some parsing error on busybox_1.0.bb May 26 15:28:54 proti_: svn up bitbake May 26 15:29:23 and bk pull while you're at it May 26 15:29:28 reenoo_: When was it fixed ? A few minutes ago ? May 26 15:29:43 yesterday May 26 15:30:06 svn update May 26 15:30:07 a few minutes before the patch in question was applied to bitbake_1.00.bb. May 26 15:30:16 at revision 223 May 26 15:30:41 I synch'ed everything 3 hours ago. May 26 15:30:53 pb_: s/bitbake/busybox/? :) May 26 15:31:03 er, yeah May 26 15:31:07 busybox_1.00.bb May 26 15:32:05 Ok, bk was not up to date. May 26 15:32:39 proti_: what would be your stragey to find that a cache is outdatet? May 26 15:33:40 file is newer, or parents in cache or file are newer. May 26 15:35:53 zecke: What do you have in mind ? May 26 15:37:45 Hum, bitbake does not like this line : 'pkg_postinst_${PN} () {' in busybox_1.00.bb May 26 15:39:36 proti_: your bitbake isn't up to date May 26 15:40:18 reenoo: What rev is bitbake ? I have 223. May 26 15:44:53 g'night May 26 15:44:56 proti_: sounds like you're running another (outdated) copy instead of what you've checked out. May 26 15:45:04 'night mickeyl May 26 15:45:13 don't forget to always rm -rf tmp/cache after a bitbake upgrade May 26 15:45:57 heh, right. forgot about that. May 26 15:45:58 mickey|zzZZzz: night. I did remove cache, but no luck for this one. Since there is a parsing error, the file cache is missing. May 26 15:46:47 reenoo_: Outdated, how ? May 26 15:47:49 proti_: wrong path or wrong BBPATH May 26 15:48:02 proti_: ok something smarter than checking every file May 26 15:51:51 zecke: You Need some heuristic to make the cache smarter and use the bitbake semantic to remove false positive. May 26 15:52:24 ok, fixed. was a system installed bitbake core that was interfering. May 26 15:52:40 My mistake. thank you. May 26 15:52:49 ahh, damn May 26 15:52:53 ~lart NLS May 26 15:52:54 * ibot eats NLS and falls over dead May 26 15:53:30 someone wants to look at the VC_svn.pm? May 26 15:53:30 I hate that lart May 26 15:53:59 zecke: What is it ? May 26 15:54:34 is there some issue with quilt_native currently? May 26 15:55:36 zecke: tinderbox didn't have a VC_svn already? Strange May 26 15:55:46 I'll look at it May 26 15:55:53 ok night all May 26 16:01:28 zecke: you just going to pull the tarball snapshots instead of bk? May 26 16:02:37 treke: currently I set up tinderbox for bitbake and running bitbake tests May 26 16:02:48 treke: I plan to use the NWL bk version May 26 16:03:13 viewcvs integration is now missing May 26 16:03:53 CosmicPenguin: how do I catch an exception? May 26 16:06:00 zecke: you mean like a signal handler? May 26 16:06:10 CosmicPenguin: handhelds.org/~zecke/VC_svn.pm May 26 16:06:20 CosmicPenguin: I dunno how it is called in perl :} May 26 16:06:38 CosmicPenguin: If I use an invalid svn revision (a not yet existing one) May 26 16:06:49 CosmicPenguin: I get some sort of error which stops tinder.cgi May 26 16:07:04 I would prefer to ignore this error May 26 16:08:13 zecke: so it dies? May 26 16:09:19 Your best best is to use eval May 26 16:14:17 zecke: eval { block } - if it fails, then the string $@ will tell you what happened May 26 16:14:22 if $@ is zero length, it was successful May 26 16:14:50 CosmicPenguin: how do I test for zero? using scalar? May 26 16:15:41 length($@) will work May 26 16:15:50 eval { May 26 16:15:50 $a = 10; $b = 0; May 26 16:15:50 $c = $a / $b; May 26 16:15:50 }; May 26 16:15:50 print "Error\n" if length($@); May 26 16:15:51 print "Done\n"; May 26 16:15:56 There's a small example May 26 16:16:54 Its ugly - Python's exception handling is far superior May 26 16:20:44 * CosmicPenguin sneaks away quitely before he screws up his uclibc build again May 26 16:21:35 hehe May 26 16:21:42 hmm eval does not do the trick May 26 16:35:30 ok tomorrow I will set up mail and write my tinderclient May 26 16:35:51 and then I'll try to migrate that setup to some server at hh.org france can (hopefully) spare May 26 16:42:47 zecke: I have been working all day on the servers May 26 16:42:53 zecke: getting closer.... May 26 16:45:12 france: I'm slowly progressing as well, I can display svn commits and now are familiar with that part of the tinderbox May 26 16:46:00 france: adding BitKeeper support (using no whining client) will be easier May 26 16:46:58 france: did you start with your port yet? May 26 16:47:23 yup, I am working on the kernel May 26 16:53:12 france: my requirements are currently low May 26 16:53:30 france: diskspace, ~6 email addresses, bandwidth May 26 16:53:46 france: if we decice to run tests at hh.org as well we might need more computing power as well May 26 18:25:17 evening May 26 18:27:03 is anyone else having problems w/ configure b0rking in "checking for bison..."? I've had this happen in two different apps (gftp and popt) May 26 18:51:37 err May 26 18:52:02 +mhentges@oe-head:..p/staging/i686-linux/bin >cat bison May 26 18:52:02 #!/bin/sh May 26 18:52:02 exec /home/mhentges/OpenEmbedded/tmp/staging/i686-linux/bin/bison -y "$@" May 26 18:52:12 what the hell? May 26 18:56:43 Well, bison-native is rebuilding, so I'll find out soon May 26 18:57:48 that makes sense to me May 26 18:58:38 I've found the problem, yacc was symlinked to bison May 26 18:59:28 probably from an earlier version of bison, there were changes recently May 26 18:59:55 yep, removing the symlink fixes the problem May 26 19:55:39 so word of warning May 26 19:56:00 apparantly it's in bad taste to send a chopstick flying across a nice chinese restraunt May 26 19:56:32 *twitch* i hate kernel oopses May 26 19:57:17 treke: did you at least engineer a quality launching device? May 26 20:02:45 is it normal for OZ to oops loading the orinoco_cs module :mad: May 26 20:04:19 yes ;) May 26 20:05:03 Why :( would it work if i built a new kernel/module with OE? May 26 20:09:44 i should go back to 3.2 and pretend 3.5 never existed May 26 20:10:04 df00z: this is not a distribution support channel. May 26 20:10:12 Rite I know. May 26 20:10:36 PS, was OE's build system ever fixed? Last time I used it 512MB was not enough May 26 20:10:56 Of ram, that is May 26 20:28:59 it was broken? May 26 20:33:43 treke: it uses a crapload of ram May 26 20:33:51 Err, used. I haven't used it since. May 26 20:35:39 it did so by design :p May 26 20:38:48 it has been improved though. May 26 21:26:27 df00z: yes **** BEGIN LOGGING AT Thu May 26 23:17:54 2005 **** ENDING LOGGING AT Thu May 26 23:59:56 2005