**** BEGIN LOGGING AT Thu Jun 09 23:59:57 2005 Jun 10 00:36:40 morning folks Jun 10 00:47:01 imorning all Jun 10 00:47:07 RP moin! Jun 10 00:51:34 RP morning Jun 10 00:53:37 RP: NWFPE fixed? Jun 10 00:54:03 do13: In theory. I still need to test that patch :) Jun 10 00:54:27 do13: It certainly removes the things Russell was complaining about... Jun 10 00:54:41 Another cadidate for "checked by: Richard Purdie"? Jun 10 00:55:00 do13: No doubt :-/ Jun 10 00:57:42 anyone of you using the new CSL 3.4.3 gcc with Qt? Jun 10 00:59:55 schurig: After all the problems I saw with the csl toolchain, I've just been using the normal one... Jun 10 01:02:58 RP: packages/gcc/gcc-cross*3.4.4 is a better advice? Jun 10 01:03:33 RP: or even the "good old 3.4.3" ? Jun 10 01:03:42 schurig: I think I still use 3.4.3 since I hardcoded it in local.conf... Jun 10 01:04:40 schurig: Its just at the moment, toolchain issues are not something I need :) Jun 10 01:14:00 03koen 07 * r1.3512.1.1 10openembedded/packages/xserver/xserver-kdrive_cvs.bb: xserver-kdrive_cvs.bb: refresh patches Jun 10 01:25:30 03koen 07 * r1.3519 10openembedded/packages/xextensions/xextensions_20050610.bb: Add xextensions snapshot to match the kdrive snapshot Jun 10 01:25:32 03koen 07 * r1.3518 10openembedded/packages/xserver/xserver-kdrive_20050610.bb: Add new kdrive snapshot, requires a new xextensions too Jun 10 01:30:37 This 2.6.12-rc5 kernel feels quite fast on the C760... Jun 10 01:32:09 RP, faster than 2.4 or faster than 2.6.11 ? Jun 10 01:33:18 jacques: It seems one of the more responsive 2.6.x kernels I've tried Jun 10 01:35:09 since you've fixed all the bugs, maybe we should move to it for openslug Jun 10 01:35:34 ixp420 Jun 10 01:36:08 jacques: I've mainly fixed pxa bugs but 2.6 does seem quite happy Jun 10 01:36:55 I was wondering how many of those bugs affect ixp - I know it's not pxa but it is xscale Jun 10 01:38:07 jacques: In that case you want the copypage fix I recently worked out... Jun 10 01:38:52 aha :-D Jun 10 01:38:54 03koen 07 * r1.3520 10openembedded/packages/xserver/xserver-kdrive_20050207.bb: xserver-kdrive_20050207.bb: tweak PV, do 'rm x*20050207*.ipk' in your deploydir/feeds to unconfuse ipkg-imake-index -m Jun 10 01:39:29 koen|afk: If I built a gpe-image yesterday with the lastes xserver, is it going to work? Jun 10 01:39:43 we're currently on 2.6.11.2 Jun 10 01:39:59 RP: no 20050207 > 0.0cvs for ipkg Jun 10 01:40:07 good morning all Jun 10 01:40:28 morning koen :) Jun 10 01:40:59 morning koen Jun 10 01:41:05 hey jacques Jun 10 01:43:37 03koen 07 * r1.3521 10openembedded/conf/distro/preferred-gpe-versions.inc: preferred-gpe-versions.inc: reflect xserver changes Jun 10 01:43:38 koen: Can I force it to use cvs somehow? Jun 10 01:43:55 RP: bk pull and do that rm command Jun 10 01:45:06 or do something line PREFERRED_VERSION_xserver-kdrive="0.0cvs${CVSDATE}" Jun 10 01:45:33 the latter will blow up with openzaurus-3.5.4.conf for obvious reasons Jun 10 01:47:59 RP: has the copypage bug introduced after 2.6.11 ? Jun 10 01:48:25 schurig: It was introduced in 2.6.12-rc4-git1 Jun 10 01:48:38 RP: ok, then I'm on the save side :-) Jun 10 01:48:55 If it had gone into 2.6.12 it would have upset a lot of people... Jun 10 01:50:01 RP: naaaah, everybody uses a lubbock Jun 10 01:50:13 and RMK *can't* be wrong Jun 10 01:53:45 ~lart rmk Jun 10 01:53:46 * ibot moos at rmk Jun 10 01:53:55 koen: although he does admit to forgetting the existance of the pxa/xscale :) Jun 10 01:56:05 hey bluelightning Jun 10 01:56:05 grrrrrrr Jun 10 01:56:10 hi all Jun 10 01:56:13 hi koen Jun 10 01:57:04 didn't you get the memo? rmk == god Jun 10 02:00:14 morning Jun 10 02:10:17 jacques: How's the strace going ? Jun 10 02:10:42 morning chaps Jun 10 02:10:49 hi mickey|tv Jun 10 02:11:03 morning pb_ Jun 10 02:11:04 good morning Jun 10 02:11:45 morning mickey Jun 10 02:13:11 pb_: morning Jun 10 02:15:36 proti, after I found a working version (4.5.9) I didn't look at it much more - I did email the strace-devel list tho - hmm I forgot to email that redhat guy Jun 10 02:17:21 jacques: Could isolate the culprit patch that breaks. 9 -> 10 touches mips and alpha (no incidence for you). There is 2 or 3 hunk that can be responsible for breakage. Jun 10 02:20:31 03mickeyl * r248 10bitbake/lib/bb/shell.py: Jun 10 02:20:31 - capture bb.parse exception in 'filebuild' Jun 10 02:20:31 - restore sys.stdout during raw_input, this repairs Jun 10 02:20:31 tab completion which I broke two revisions ago Jun 10 02:21:01 pb_: would you have some time perhaps today to add the h1940 to tslib? I'd be really grateful Jun 10 02:24:24 Arjan: I'll take a look. Jun 10 02:24:36 pb_: thanks! Jun 10 02:25:44 jacques: following from the lament we had the other day about the armeb patches in gcc, do you know who is the author of the uclibc patches in binutils? Jun 10 02:28:24 dang, I used to know that... Jun 10 02:29:11 I'm just adding binutils 2.16, and none of the damn uclibc patches apply any more. Jun 10 02:29:48 any help in #uclibc ? Jun 10 02:29:51 Updating them to work with the new version is a fairly easy mechanical exercise, but it doesn't seem like the kind of thing that ought to be necessary. Whoever wrote those patches should get them checked in to CVS. Jun 10 02:31:41 very true Jun 10 02:35:22 03koen 07 * r1.3522 10openembedded/conf/distro/familiar-0.8.3.conf: Rename: conf/distro/familiar-0.9.0.conf -> conf/distro/familiar-0.8.3.conf Jun 10 02:51:24 anybody messed with the Virtera arm simulator? they say it can emulate a Z 5500 and boot opie Jun 10 03:12:18 today is really dragging alone :( Jun 10 03:12:31 s/alone/along Jun 10 03:13:04 ade|desk: hmm... mine didn't Jun 10 03:13:33 power cut at work early this morning, and troubleshooting nasty bugs for most of the rest of the day Jun 10 03:21:25 ih! Jun 10 03:21:27 err. hi! Jun 10 03:21:28 bluelightning: we're getting a nice new generator here soon, so even powercuts will not slow us down ... where is the fun in that ? Jun 10 03:21:52 heh Jun 10 03:22:26 our UPSs had just enough capacity to cover our servers for the outage Jun 10 03:22:40 if it had been a minute longer they would have gone down Jun 10 03:23:04 03pb 07 * r1.3521.1.1 10openembedded/packages/binutils/ (2 files in 2 dirs): add binutils-2.16; set DEFAULT_PREFERENCE=-1 to linux-uclibc due to bit-rotten patches Jun 10 03:23:04 fortunately long power cuts don Jun 10 03:23:32 't happen too often there, but we get micro-outages about every 2 months Jun 10 03:23:56 I was trying to make a new package. I made a new directory under 'packages', made a .bb file, started bitbake -i, did a 'parse'.. but when I try building it it always says 'ERROR: Nothing provides package" .. is there some way to debug that? Jun 10 03:24:10 bluelightning: we discovered that our UPS batteries were shagged ~january, only had them refurb'ed 6 months earlier Jun 10 03:24:31 damn :( Jun 10 03:25:18 ... _ is not allowed in package names? Jun 10 03:25:24 patw: correct Jun 10 03:25:29 03pb 07 * r1.3524 10openembedded/packages/binutils/ (binutils.inc binutils-cross.inc binutils-cross_2.16.bb): forgotten files from last changeset Jun 10 03:25:29 rather than replace the compaq UPSs we are putting in a gen and an uber ups to hold out for the gen to kick in :) Jun 10 03:25:36 good to know. now it works :) Jun 10 03:26:05 pb_: you got uclibc patches in now then ? Jun 10 03:26:11 no Jun 10 03:26:58 is there anything else I need to know about package names? Jun 10 03:27:02 anything else that isn't allowed? Jun 10 03:27:25 I think the bitbake manual should tell you what's allowed. Jun 10 03:28:04 I wouldn't call it a manual. anyway, thanks. Jun 10 03:28:09 03pb 07 * r1.3525 10openembedded/packages/binutils/binutils_csl-arm-20050603.bb: update binutils csl-arm to new snapshot Jun 10 03:28:52 pb_: any news on the eabi glibc bits? Jun 10 03:28:56 no Jun 10 03:29:32 right now, the HEAD doesn't even compile for the old abi, so I haven't been worrying too much about the eabi yet. Jun 10 03:29:44 heh Jun 10 03:35:49 03koen 07 * r1.3524 10openembedded/packages/base-files/base-files/h3900/fstab: comment mtd2 for h3900 to unbreak h51xx ipaqs Jun 10 03:35:52 03koen 07 * r1.3523 10openembedded/packages/maemo/ (sapwood/cflags.patch sapwood_1.20-1.bb): Add sapwood, part of the maemo.org suite, courtesy Florian Boor Jun 10 03:38:32 03koen 07 * r1.3526 10openembedded/packages/base-files/base-files_3.0.14.bb: base-files_3.0.14.bb: and bump PR Jun 10 04:59:35 hi all Jun 10 05:01:07 I'm trying to complile links for x11 but I'm getting the following error. /stuff/tmp/cross/lib/gcc/arm-linux/3.4.3/../../../../arm-linux/bin/ld: cannot find -lfl Jun 10 05:02:32 darmou, that's part of flex Jun 10 05:02:36 AFAICT Jun 10 05:02:42 flex cool thanks Jun 10 05:03:32 might I suggest you look at elinks? it's superior to links in just about every way I think Jun 10 05:03:40 * darmou adds flex to the depends list Jun 10 05:04:08 ok thanks for the tip, I've also tried to compile minimo but the patch breaks with the latest cvs Jun 10 05:04:17 BTW, I'm not sure but it may be flex-dev if there is such a thing Jun 10 05:05:23 yeah libfl.a is in flex-dev in oe - I expected to find a shared lib but don't see one Jun 10 05:06:08 hmm my desktop doesn't have a shared one either Jun 10 05:06:40 ERROR: Nothing provides flex-dev Jun 10 05:07:04 if you bitbake flex, it makes several packages, one of which is flex-dev Jun 10 05:07:26 03mickeyl 07 * r1.3527 10openembedded/packages/flex/flex_2.5.31.bb: flex: install libflex into STAGING Jun 10 05:07:35 heh Jun 10 05:07:37 rebuild flex, then rebuild links Jun 10 05:07:41 ~lart hrw Jun 10 05:07:44 * ibot slaps a compatible dib on hrw's head Jun 10 05:07:48 checking minimo as we speak Jun 10 05:07:54 darmou, looks like you found a bug :-) Jun 10 05:08:26 looks like flex is built however links still does not built, gets the same error:( Jun 10 05:08:41 ok I'll try rebuild flex Jun 10 05:09:00 hrw recently removed the staging code from the .inc Jun 10 05:09:06 darmou, as mickeyl said, rebuild flex, then rebuild links (after pulling the latest) Jun 10 05:09:16 hence libflex didn't make it into staging hence you get that linking error Jun 10 05:10:10 heh Jun 10 05:10:12 why did he do that? Jun 10 05:10:19 how do you rebuild a package? Jun 10 05:10:48 probably because I verbosely copied flex-native's do_stage() into the .inc which was a bit over the top, i.e. install flex ${STAGING_BINDIR}/ ... Jun 10 05:15:13 darmou: NOTE: package minimo-0.0cvs20050610-r8: task do_patch: completed Jun 10 05:15:40 great thanks koen Jun 10 05:15:59 does bitbake -f flex rebuild the package? Jun 10 05:16:18 doh aparently not sigh Jun 10 05:16:21 bitbake -b path/to/flex.bb -c clean Jun 10 05:16:25 bitbake -b path/to/flex.bb Jun 10 05:16:39 bbnew () Jun 10 05:16:39 { Jun 10 05:16:39 bitbake -c clean -b $1 && bitbake $BBD -b $1 Jun 10 05:16:39 } Jun 10 05:17:08 or 'rebuild flex' from bb -i Jun 10 05:17:10 :D Jun 10 05:17:13 groovy Jun 10 05:17:19 excellent thanks koen/mickeyl Jun 10 05:18:07 * koen wonders why the h3600 gpe-image is failing Jun 10 05:21:27 aha Jun 10 05:22:21 03koen 07 * r1.3526.1.1 10openembedded/conf/distro/preferred-gpe-versions.inc: preferred-gpe-versions.inc: prefer the xextenstion 20050610 snapshot Jun 10 05:22:34 excellent Jun 10 05:23:00 I've elected my good 'ole trusty 3650 to feature a permanent gpe-image installation from this weekend onwards Jun 10 05:23:29 * koen looks at the opie screen on the hx4700 Jun 10 05:24:10 mickeyl: What is the plan with the shell ? An extension ? or will it become the heart of bitbake ? Jun 10 05:24:13 * koen tries to flash qtopia 2.1.1 Jun 10 05:25:50 damn still getting the same link error, even after rebuilding both flex and links:( Jun 10 05:26:17 did you bk pull in between? Jun 10 05:26:18 proti: The shell is more or less just an additional way to communicate with the bitbake core. The bitbake command line utility will stay like it is, however the shell will grow more sophisticated. Jun 10 05:26:45 proti: long term plan is once we know what features we need to restart with some graphic interface Jun 10 05:27:16 the good thing is that the shell helps to motivate us to refactor things in the bb core Jun 10 05:27:41 ahh ok thanks koen Jun 10 05:28:13 there are also quite some more performance tweaks in sight, i.e. through the use of generator comprehensions Jun 10 05:28:31 s/comprehensions/expressions/ Jun 10 05:28:51 python 2.4 only though Jun 10 05:29:11 mickeyl: I would be a great help to drop to interactive mode whenever an error occurs. Jun 10 05:29:40 proti: oh yeah, that sounds good Jun 10 05:29:41 I thought oe was moving to subversion? Jun 10 05:29:57 hell no Jun 10 05:29:59 darmou: who told you that? Jun 10 05:30:03 mickeyl: where is this shell, lib/bb/shell.py ? Jun 10 05:30:15 schurig: bitbake -i Jun 10 05:30:17 schurig: yes. you don't invoke it that way though. 'bitbake -i' Jun 10 05:30:20 what is oe moving to? Jun 10 05:30:28 it's not decided yet. Jun 10 05:30:31 nothing at the moment Jun 10 05:30:31 jacques: no one knows Jun 10 05:30:34 ah Jun 10 05:30:52 jacques: it get's decided by one person in a rush on 1st july, when bkbits.net will be switched off :-) Jun 10 05:30:57 http://oe.handhelds.org/ Jun 10 05:31:11 schurig, lol let's hope not :-) Jun 10 05:31:20 schurig: 20 minutes to pull monotone with monotone here. Jun 10 05:31:22 Quote from that web page "We also decided that the core didn't really benefit from the peer to peer nature of BitKeeper, so we moved the core to Subversion over at berliOS." Jun 10 05:31:32 darmou, that refers to bitbake Jun 10 05:31:34 i don't think that's too bad considering the amount of history in the monotone tree Jun 10 05:31:44 What about git then ? Following linus path... Jun 10 05:31:51 darmou: OE is not moved (with metadata) but bitbake as executable yes... Jun 10 05:31:53 git can't push Jun 10 05:32:11 proti: add "cg-push" to cogito, and we'll use cogito/git :-) Jun 10 05:32:32 I went to a linux talk on baz and baz-ng that was quite interesting Jun 10 05:32:35 a web frontend that does more that link to kernel.org Jun 10 05:34:06 schurig: cg-commit ? Jun 10 05:34:20 proti: this commits a change into your local .git repository Jun 10 05:34:25 proti: just like bk commit does Jun 10 05:35:20 schurig: Then how kernel people are doing work ? They export a diff and send them to linus ? Jun 10 05:35:43 proti: no, they make their .git available via rsync and as Linus to pull from them Jun 10 05:35:47 linus pulls in their branch or imports a patch Jun 10 05:36:01 s/as/ask/ Jun 10 05:36:24 That is basicaly what a git-push should do. Jun 10 05:37:06 proti: In a git push, you'd have to tell the site hosting the public .git tree to pull from you Jun 10 05:37:08 ok did the bk pull in between still get same error Jun 10 05:37:55 proti: so the site hosting the public rsync tree needs some way to know about you, your identity/right. And you need to export your .git subdirs via rsync as well. Not that nice. Jun 10 05:38:15 proti: maybe exporting a patch and injecting it into the public git via ssh is better Jun 10 05:38:21 mickeyl: could you sync? Jun 10 05:38:21 whatever, no one wrote it yet Jun 10 05:38:26 schurig: rsync over ssh with public keys ? Jun 10 05:39:02 that would limit the merging ability Jun 10 05:39:04 proti: AFAIK it's more than rsync. If you just rsync your .git tree and you break the connection in the middle, then the repository would be in an inconsistent state Jun 10 05:39:34 anyway, I stop now with that, I'm not deep enougth into git (and neither are you) to really know what would be necessary for a sane implementation :-) Jun 10 05:40:12 Indeed. I need to practice a litlle with git to understand the drawbacks. Jun 10 05:40:48 it might be easier to use something different than git then, mickeyl is for Monotone Jun 10 05:41:05 mickeyl: btw, looks like they made this faster ... or do you have a VERY fast machine? Jun 10 05:41:56 mickeyl: All the basic command given to bitbake could be translated into one line shell commands or small combination of command (in scripts) and submitted to the interactive mode (not so interactive in this case). Jun 10 05:42:02 and git can't import the OE repository at once :( Jun 10 05:42:51 Main problem is how to migrate from OE without loosing history and comments. Jun 10 05:42:51 koen: I wouldn't mind if the current changeset history of BK would be lost Jun 10 05:43:04 schurig: no, just a Pentium 2.4Ghz here Jun 10 05:43:06 schurig: just the tree, no metadata Jun 10 05:43:40 i think we definitly wil llose hisotry Jun 10 05:43:42 cogita is shell, and can't handle cg-add file1 file2 ...file2500 Jun 10 05:44:02 koen: find -type f | xargs cg-add Jun 10 05:44:20 we can use a read-only cvs repository to review history before our move Jun 10 05:44:21 I'd rather want cogite-add to work Jun 10 05:44:38 koen: go forward, it's an open source project :-) Jun 10 05:44:48 cvs2svn, svn2monotone Jun 10 05:44:52 schurig: I tried Jun 10 05:45:02 koen: but your name isn't Pasky :-) Jun 10 05:45:02 but cogito is to obfuscated Jun 10 05:45:10 Did bitmover say that it will convert and left trees to cvs repo ? Jun 10 05:45:16 /and/any/ Jun 10 05:45:56 proti, yes they offer this to any project Jun 10 05:46:01 anyway, I'll be incommunicado from 20050701 to 20050717, so I'll see what happens Jun 10 05:46:09 koen: vacation? Jun 10 05:46:12 yes Jun 10 05:46:15 where ? Jun 10 05:46:23 florida and new orleans Jun 10 05:46:28 woah nice Jun 10 05:46:53 * mickeyl wanders home Jun 10 06:01:01 03koen 07 * r1.3529 10openembedded/packages/linux-hotplug/files/mmc.agent: mmc.agent: add sleep 1 to mmc.agent to work around buggy drivers Jun 10 06:05:27 koen: That doesn't sound nice :-( Jun 10 06:05:54 Does anyone mind if I upgrade oprofile to version 0.9? Jun 10 06:06:53 I might also add a cvs version... Jun 10 06:06:56 RP: it doesn't, but at least sd is working on the hx4700 100% of the times now Jun 10 06:07:29 koen, what sort of SD read/write speeds on the hx4700 ? Jun 10 06:07:29 koen: That's on the hx4700?! That's bad :-( Jun 10 06:07:59 jacques: haven't really benchmarked it yet Jun 10 06:08:27 trying to figure out why bootldr doesn't flash stuff correctly Jun 10 06:08:27 koen, I'd love to know if you get a chance Jun 10 06:08:42 but obviously you have bigger fish to fry right now :-) Jun 10 06:08:44 will let you know Jun 10 06:25:38 hey prpplague Jun 10 06:25:45 koen: morning Jun 10 06:25:55 koen: whats cookin today? Jun 10 06:26:27 lot's of tedious matlab code Jun 10 06:27:07 koen: lovely Jun 10 06:27:26 * prpplague gave birth (because it was that painful) to a lcd driver yesterday Jun 10 06:31:53 prpplague: pl100? Jun 10 06:35:17 ade|desk: pl110 yea Jun 10 06:35:37 groovy Jun 10 06:36:11 prpplague: barbados.org says you are at 14.4N, 54.4W Jun 10 06:36:31 koen: sounds right Jun 10 06:37:06 prpplague: according to my maps (keyhole) that's 330 miles off Jun 10 06:37:17 * prpplague looks Jun 10 06:37:56 13'06N 59'37W should be where bridgetown is Jun 10 06:38:19 yea thats what i got as well Jun 10 06:38:30 koen: hehe Jun 10 06:38:55 koen: http://www.cia.gov/cia/publications/factbook/geos/bb.html Jun 10 06:39:51 ah yes, the cia doesn't direct me into the big puddle Jun 10 06:48:09 koen: if you look at the map there on the cia site, my house in on the coast line just south of the airport Jun 10 06:48:33 koen: and my offices are right near "the crane" Jun 10 06:52:14 hi prpplague Jun 10 06:52:29 pb_: morning Jun 10 06:52:37 pb_: hows things in Histon today? Jun 10 06:52:57 pb_: hehe, heading out for a friday night pub crawl? Jun 10 06:53:06 prpplague: seemed good enough when I left there this morning Jun 10 06:53:27 pb_: hehe, does that mean you are someone else today? Jun 10 06:53:34 pb_: Jun 10 06:53:43 prpplague: well, Histon is where I live. right now I'm at our office, which is on the other side of town. Jun 10 06:53:55 pb_: ahh, gotcha Jun 10 06:54:18 pb_: wasn't sure about how big Histon is Jun 10 06:55:15 it's a medium-size village, maybe a thousand people or so. Jun 10 06:55:23 pb_: ahh Jun 10 06:55:38 pb_: hehe did you see my messages on elinux about my pl110 problem? Jun 10 06:56:12 pb_: found a nice english style pub on the north side of the island last weekend Jun 10 06:56:27 pb_: good ale and shepards pie :) Jun 10 06:56:31 see http://www.streetmap.co.uk/newmap.srf?x=544245&y=264047&z=3&sv=544245,264047&st=4&mapp=newmap.srf&searchp=newsearch.srf&ax=544245&ay=264047 or some such Jun 10 06:56:46 our offices are about where it says "Church End" on the southeast side of Cambridge. Jun 10 06:56:59 prpplague: ah, sounds cool Jun 10 06:57:18 no, I didn't see your note about the pl110. did you fix your problem? Jun 10 06:58:09 pb_: the sharp docs seemed to have left out a little fact that one of the registers bits turns of all the data lines to the lcd Jun 10 06:58:27 pb_: and the primecell docs only mentioned it in a footnote Jun 10 06:58:37 s/of/off Jun 10 07:03:41 prpplague: hah! Jun 10 07:03:55 I can see how that could spoil your day. Jun 10 07:06:24 pb_: yea, i've been banging my head on it for almost two weeks Jun 10 07:06:37 pb_: i kept thinking i had something wrong with my software Jun 10 07:17:29 yah, I hate it when that happens Jun 10 07:29:19 03pb 07 * r1.3530 10openembedded/packages/binutils/binutils-cross_csl-arm-20050603.bb: add -cross version of latest binutils, courtesy bero@arklinux.org, per bug #76 Jun 10 07:36:29 hi Jun 10 07:37:22 hey CoreDump|home Jun 10 08:00:37 have a good weekend everybody Jun 10 08:03:47 Hi CoreDump Jun 10 08:04:06 I'm trying to complile links but get this error message ld: cannot find -lfl Jun 10 08:04:29 I've been told that it's looking for flex, however I have built flex and still get that error message Jun 10 08:04:41 Could there be something else it may be looking for? Jun 10 08:11:59 darmou: did you build flex-2.5.31-r3 ? Jun 10 08:17:01 ahh /tmp/work reports flex-2.5.31-r0 Jun 10 08:17:53 CoreDump|home: grabbed a hentges image and played with it. I like the altboot strategy and I want us to think about how to make this the default for OZ Jun 10 08:18:18 CoreDump|home: i would like to have an automated timeout in the boot menu though Jun 10 08:18:31 e.g. 5 seconds and then the default option will be chosen Jun 10 08:18:35 mickeyl: namepicking, vote, soon Jun 10 08:18:35 darmou, sounds like you didn't get the latest changesets Jun 10 08:18:44 mickeyl: I tried that w/ the little program used by init.d/opie but it was a no-go Jun 10 08:19:31 the problem was that the "any" key you press at the prompt will be read by busybox's "read" command so it fubars the menu Jun 10 08:19:40 CoreDump|home: heh Jun 10 08:19:43 :) Jun 10 08:19:45 i'll think about that Jun 10 08:19:56 i definitly want a boot menu for oz 3.5.4 Jun 10 08:20:35 I agree that a timer is a must-have if you decide to put it into the images Jun 10 08:21:04 can you comment on what's needed to make CF and SD boot working? sounds like you have quite a lot of customized packages in your image Jun 10 08:21:10 what's all the hentges-utils ? Jun 10 08:21:35 mickeyl: altboot is in OE and doesn't require any special hacks to work. Just a config file Jun 10 08:22:19 http://oz.hentges.net/openembedded/daily-snapshot/2005/June/hentgescvs_10-Jun-2005.tar.gz Jun 10 08:23:44 morning Jun 10 08:24:07 morning chouimat Jun 10 08:24:47 the only problem I see currently would be that someone needs to sit down and work out the steps needed to get CF and SD support working on clam-shell kernel 2.6 Zaurii Jun 10 08:25:06 it's trivial but someone needs to do it :) Jun 10 08:25:54 righto Jun 10 08:26:46 however, i think this shouldn't happen before the pcmcia rework has been finished Jun 10 08:27:00 too much variables atm. Jun 10 08:28:29 mickeyl: it might even not be OZ anymore :) Jun 10 08:28:38 *nod* Jun 10 08:28:59 * RP should get around to testing pcmciautils at some point Jun 10 08:29:53 any C3K enthusiasts in da house ? Jun 10 08:30:08 i want to get rid of pivot_boot for the C3K images. Jun 10 08:30:25 i have a much better feeling when we just mount the harddisk as /usr Jun 10 08:30:36 any objections to that ? Jun 10 08:30:46 err yeah, /opt would have to be linked Jun 10 08:30:48 :D Jun 10 08:31:59 hi dirk Jun 10 08:32:07 hi mickeyl Jun 10 08:32:29 hmm i guess noodles was the only other C3K owner in #oe Jun 10 08:32:34 that means no objections to my plans :D Jun 10 08:32:37 :) Jun 10 08:33:35 mickeyl: :) Jun 10 08:33:44 but symlinking is going to be a pain. You'd need a bare-bone /usr inside the ROM for it to boot at all, then mount -bind the real user on top of it Jun 10 08:34:02 s/user/\/usr/ Jun 10 08:36:16 * kergoth hugs bind mounts Jun 10 08:36:36 mickeyl: switch to {bindir} Jun 10 08:37:35 hmm, I need to figure out all this mounting stutt Jun 10 08:38:13 I need to put basic stuff in flash and overlay my microdrive for bulky stuff and writable stuff Jun 10 08:38:21 bind mounts are basically like normal mounts, except that instead of mounting a device at a mountpoint, you're mounting another directory there. Jun 10 08:38:32 ah Jun 10 08:39:11 so that isn't quite what I want, but maybe it is :) Jun 10 08:39:12 like multiple chroots sharing the same /var Jun 10 08:39:28 kergoth: played with unionfs lately? Jun 10 08:40:22 not recently. i havent been very active in open source lately. now that i'm mostly settled in mn, and have my stuff and all, will probably do some work soon Jun 10 08:40:23 heh Jun 10 08:40:42 apt-get remove WoW Jun 10 08:41:23 hehe Jun 10 08:41:43 kergoth: that sounds good (bind mounts) Jun 10 08:42:02 koen: i can't. a whole of stuff would break if we do that now :/ Jun 10 08:42:08 (opt -> bindir) Jun 10 08:42:18 besides... ljp would get a heart attack Jun 10 08:42:20 we don't want that Jun 10 08:42:21 heh Jun 10 08:43:08 too bad /opt is actually valid in FHS Jun 10 08:43:31 yah Jun 10 08:43:48 I guess ljp will say f.d.o is bullshit for pdas Jun 10 08:44:18 nah, he would say "fluff" Jun 10 08:44:21 :D Jun 10 08:45:59 wb ibot Jun 10 08:46:13 I'm curious mickeyl. Why don't you like the pivot_root boot? IMO It is a pretty perfect solution for the 3000 and its large microdrive. Jun 10 08:46:29 just screw the tiny ROM :) Jun 10 08:47:14 powerconsumption and access/seek time Jun 10 08:47:36 CoreDump|home: a) a bootstrap image fits perfectly in that flash. b) the pivot_root boot sequence is slooooooow c) the pivot_root is more fragile than i want it to be given that the hd is registered as slot1 Jun 10 08:47:38 ~lart sharp Jun 10 08:47:38 * ibot does a little 'renice 20 -u sharp' Jun 10 08:48:01 stick a CF card into and boom Jun 10 08:48:04 no booty Jun 10 08:48:11 * mickeyl shakes head Jun 10 08:48:14 eh Jun 10 08:48:18 really? lol Jun 10 08:49:10 nothing that we can't work around with some filesystem checks, but it's ugly. i don't like workarounds Jun 10 08:49:20 mounting it as /usr would be just more natural Jun 10 08:49:35 or one partition as /usr and the other as /opt Jun 10 08:49:37 *shrug* Jun 10 08:50:00 well, you could mount the HD by UUID or volume label. Jun 10 08:52:10 true enough. i'd rather make it a bit more sane though :) Jun 10 08:52:25 * mickeyl looks at b0ti Jun 10 08:52:32 proti: ping Jun 10 08:53:23 mickeyl: did you look at that ARM Emulator yet? Jun 10 08:53:53 zecke: no chance yet Jun 10 08:54:00 did you? Jun 10 08:54:09 no Jun 10 08:55:04 mickeyl: I suspect swapping the cardslots around in the pcmcia code might have the desired effect Jun 10 08:55:36 RP: that'd be good Jun 10 08:55:55 mickeyl: The hx2750 is confused about its slot numbering as I got it wrong when I wrote the driver :) Jun 10 08:56:07 hehe, i see Jun 10 08:56:25 kergoth: Are you happy for me to upgrade oprofile to 0.9 and add a cvs.bb for it as well? Jun 10 08:57:02 03koen 07 * r1.3529.1.1 10openembedded/ (4 files in 3 dirs): sapwood_1.20-1.bb: add patch, courtesy Florian Boor Jun 10 08:58:13 Is it possible to have a machine independend module_autoload_xx in one .bb file? Jun 10 09:03:38 do13: why not? Jun 10 09:04:09 zecke: what's the syntax? Jun 10 09:04:34 machin indepen... == no use of overrides? Jun 10 09:04:48 you can use include file.inc to include definitions Jun 10 09:06:10 do13: if you mean you want a module to always be autoloaded on all machines, put it in kernel.bbclass. Jun 10 09:06:16 I think there are already some examples of that in there. Jun 10 09:08:25 pb: Can i put module_autoload_.. into the machine.conf file? Jun 10 09:09:01 Er, yeah, I guess. I wouldn't really describe it as machine independent in that case, though. Jun 10 09:09:16 pb_ I want a module always be autoloaded on ONE machine. Jun 10 09:09:47 Ah. Jun 10 09:09:59 In that case, yes. Jun 10 09:10:40 Thanks Jun 10 09:10:40 do13: it could work in machine.conf Jun 10 09:14:01 this is why I think the wince software industry is just sad: http://www.pocketgear.com/software_detail.asp?id=18243 Jun 10 09:14:23 does anyone think that's worth $15? Jun 10 09:14:50 ok got a little bit futher with building links now get this error /stuff/tmp/staging/arm-linux/lib/libpng.so: undefined reference to `deflate' Jun 10 09:15:32 darmou, progress - good Jun 10 09:15:36 sounds like libpng is getting built wrong. Jun 10 09:15:41 check that it has -lz in its final link. Jun 10 09:17:34 jacques: LOL, 15$, right Jun 10 09:18:58 CoreDump|home, exactly :-\ Jun 10 09:19:17 just browsing pocketgear makes me feel like I need a shower Jun 10 09:19:53 heh, browse the OZ feed to get your mood up again Jun 10 09:19:55 :) Jun 10 09:19:59 jacques: heh Jun 10 09:20:15 gotta wonder why you choose to browse pocketgear in that case Jun 10 09:21:08 well, my h2200 still runs wince - can't wait til I can convert it to linux full time Jun 10 09:21:35 I guess I just have massochistic tendencies Jun 10 09:22:00 ah right Jun 10 09:22:12 what functionality are you missing before you can convert it to wince? Jun 10 09:22:20 er, to linux Jun 10 09:22:42 well, last I checked which may not be totally up to date, power management wasn't working right, and no bootloader Jun 10 09:22:59 would also like decent SD perfoermance Jun 10 09:23:14 I think the bootloader thing is resolved now. Jun 10 09:23:20 yay!! Jun 10 09:23:32 aiui, you can boot linux from the HTC bootloader, including suspend/resume. Jun 10 09:23:36 pb_: using the HTC one or do we've a 100% free one? Jun 10 09:23:42 zecke: htc Jun 10 09:23:43 ok HTC Jun 10 09:23:50 yah, htc Jun 10 09:24:12 I think the ohh bootldr is nearly working, but not quite. Jun 10 09:24:14 the h2200 is just so darn convenient for using with a digital camera that writes to CF Jun 10 09:24:22 iirc, joshua destroyed his h2200 or something, so progress on that was halted. Jun 10 09:24:41 sounds about par for the course with joshua_ :-P Jun 10 09:24:47 hehe, yeah Jun 10 09:24:53 he's ... hard on hardware Jun 10 09:24:57 pb_: oh I thought he got a new mainboard Jun 10 09:25:10 cya later... Jun 10 09:25:12 it's good that he is enthusiastic about bootloaders and jtag, but his hardware does seem to have a very short life expectancy Jun 10 09:25:13 * zecke|toothache gets into bed Jun 10 09:25:18 pb_, I am out of the loop - are you saying we can boot linux using the stock bootloader on the h2200 ?? Jun 10 09:25:20 later zecke Jun 10 09:25:27 jacques: that is what I am saying Jun 10 09:25:35 wow how did I miss that??? Jun 10 09:25:39 not that I have tried it myself, mind, but I gather it works now. Jun 10 09:26:10 my downward slide came when the various list archives stopped updating Jun 10 09:26:21 that's how I used to keep current Jun 10 09:26:34 but they haven't updated since nov 24th or so Jun 10 09:26:42 yeah Jun 10 09:26:46 you can read most of them with GMANE. Jun 10 09:27:00 i thought I looked there - 2200 port list wasn't there (that I could find) Jun 10 09:27:04 ah Jun 10 09:27:17 or the other new port lists - 4700, 2700 Jun 10 09:27:23 last I checked anyway Jun 10 09:27:29 right, that's a bit sad Jun 10 09:27:36 I guess you could ask the gmane folk to add those lists. Jun 10 09:28:10 hmm, good point. also, france said he was going to fix the archives now that he was back on contract, but I gather he's been too busy with the real work Jun 10 09:28:49 yeah, it seems so Jun 10 09:29:09 ports are always more work than they first appear Jun 10 09:29:52 fwiw, mreimer sent a note to the list on May 31, saying "I just committed a patch to make suspend/resume work with the WinCE Jun 10 09:29:52 bootloader, so we now have all the pieces in place to replace WinCE with Jun 10 09:29:52 linux without the risk of bricking our h2200s, since we can keep the WinCE Jun 10 09:29:52 bootloader and use it to boot and suspend/resume." Jun 10 09:30:41 wow, that's just great. Jun 10 09:31:05 hmmm, how do we load the firmware image ? similar to hh.o bootloader? Jun 10 09:31:17 I'm not sure. I think you do it from CF somehow. Jun 10 09:31:20 probably no nice flash from CF thingi Jun 10 09:31:25 oh maybe there is then Jun 10 09:32:15 I'll definitely have to look into this - exciting Jun 10 09:32:39 03pb 07 * r1.3530.1.1 10openembedded/packages/gcc/ (gcc-cross_4.0.0.bb gcc_4.0.0.bb): add LDFLAGS patch to fix problem with f951 Jun 10 09:36:37 pb_, do you know anything about the state of ALSA on ARM 2.6 kernel ? Jun 10 09:39:21 not really, no. Jun 10 09:40:23 ok, I'll ask around Jun 10 09:40:56 jacques: the nokia people are using it Jun 10 09:40:57 03koen 07 * r1.3531.1.1 10openembedded/packages/maemo/sdk-default-icons_0.2-1.bb: Add icons for the maemo.org suite, courtesy Florian Boor Jun 10 09:41:05 the h2200 people are using it too Jun 10 09:42:24 the guy at coresound seems to think that it's such a mess that their CD digital recording card won't work on linux ARM right now Jun 10 09:42:28 wow. mkdep failed due to "argument list too long." Jun 10 09:42:34 CF, not CD Jun 10 09:42:37 it certainly used to be a big mess, but I think it's better now. Jun 10 09:42:56 are the h2200 people really using alsa? I thought they were just using the oss emulation layer on top of the alsa lowlevel drivers, same as on the other ipaqs. Jun 10 09:42:56 I was hoping his info was outdated Jun 10 09:43:33 keturn: yah, the kernel makefiles suck like that. Jun 10 09:45:22 pb_: for me alsa = alsa kernel drivers Jun 10 09:45:42 pb_: but giorgo uses aplay, so yeah, full blown alsa Jun 10 09:47:24 The PDAudio-CF Linux is stable and runs on all Linux distributions that can run ALSA other than ARM. ARM's audio package (including the relevant sound chip drivers) is in disarray and is not expected to stabilize in the near future. That precludes use of Linux PDAs, including the HP iPAQs and Dell Axims. Jun 10 09:47:29 peh. the most common answer on google seems to be "move your build directory to something with a shorter path." Jun 10 09:47:47 that's what the coresound guys said Jun 10 09:48:49 keturn: basically yes. Jun 10 09:48:56 evening Jun 10 09:49:15 it gets very long once you add in the path to staging Jun 10 09:49:45 though you can always hack mkdeps a bit Jun 10 09:49:55 that sometimes lets you get away with more Jun 10 09:50:01 jacques: you could check with ljp, he was doing some work with the PDAudio thing on the Z. Jun 10 09:51:40 pb_, thanks! Jun 10 09:52:22 it's a bit sad that it doesn't work properly on the iPAQ. I remember Len Moskowitz showing it off at one of the HDC events. Jun 10 09:53:28 pb_, yes, I saw that too, which makes me sad that after about two years it's not working Jun 10 09:53:41 03koen 07 * r1.3534 10openembedded/packages/maemo/sdk-default-theme_0.1-1.bb: Add the default theme for maemo, courtesy Florian Boor Jun 10 09:53:44 right Jun 10 09:53:56 that was the fall HDC at CRL in 2003 IIRC Jun 10 09:56:40 <[g2]> pb_ were you the guy that loaded up the icache with the xscale ? Jun 10 09:57:15 <[g2]> our program worked yesterday from Redboot loading it and directly from the mini i cache btw Jun 10 09:58:01 <[g2]> maybe that should be directly to the mini icache Jun 10 09:59:26 [g2]: I didn't implement it, but I was probably the guy that told you about the code. Jun 10 09:59:31 that sounds cool Jun 10 10:00:14 <[g2]> someone had mentioned they used the same technique on a different target Jun 10 10:00:29 <[g2]> pb_ thx again for the help on the assembler btw Jun 10 10:00:34 sure, no problem Jun 10 10:00:51 I did some work on a machine that used the icache-loading technique, but I wasn't actually the one who wrote the code for it. Jun 10 10:01:03 <[g2]> ah.. ok Jun 10 10:01:44 this thing boots that way, for example: http://www.psionteklogix.com/public.aspx?s=com&p=Products&pCat=128&pID=1082 Jun 10 10:01:59 and there's some code in handhelds.org cvs for booting an ipaq h1900 or h3900 over jtag. Jun 10 10:02:16 <[g2]> do you know if they just copied data from the icache to the dcache (after the dcache was setup obviously) for the storage space like autovars ? Jun 10 10:02:50 I think they just turned on the sdram controller and used that for variable storage. Jun 10 10:03:08 <[g2]> that's another possibitity Jun 10 10:03:36 or, yeah, you could lock down some of the dcache as RAM. Jun 10 10:03:44 <[g2]> however, I was thinking it'd be *cool* to run without flash or memory and be able to do full diagnostics Jun 10 10:03:45 has anyone been able to get a complile of beep for collie that works? Jun 10 10:03:58 yep, that would be kind of cute. Jun 10 10:04:13 <[g2]> and useful from a manufacturing perspective Jun 10 10:04:26 <[g2]> pb_ thx for info Jun 10 10:04:56 03koen 07 * r1.3535 10openembedded/packages/maemo/sdk-default-theme-config_1.0-1.bb: More maemo theme stuff, again provided by maemo-master Florian Boor Jun 10 10:06:15 gmane has 2200 port list now (yay) but AFAICT no 4700 or 2700 Jun 10 10:06:22 I've seen from the forums that someone has made some headway with finding out the buttons for gpe with collie, however it seems that the calendar and control keys are mapped to the same key and control does not work Jun 10 10:17:32 ~kill TI cardbus bridge Jun 10 10:17:33 * ibot shoots a super-inverse fluxphoton gun at TI cardbus bridge Jun 10 10:18:37 wow, MMC isn't very fast even under the best of circumstances is it? Jun 10 10:18:45 I'm getting 1.63MB/s read Jun 10 10:20:01 that's with a USB2.0 cardreader Jun 10 10:21:10 that's expected Jun 10 10:21:20 IIRC mmc has a 1bit interface Jun 10 10:21:26 so it's more or less a serial line Jun 10 10:22:34 03reenoo 07 * r1.3534.1.2 10openembedded/packages/sablevm/ (sablevm-classpath_1.11.3.bb sablevm-classpath_1.1.9.bb): sablevm-classpath: add xtst to DEPENDS. fixes bug #75. Jun 10 10:23:54 of course, putting that into perspective, that's faster than I ever got through the PCMCIA sleeve of any of my other ipaqs Jun 10 10:24:16 but far from the 5-6MB/s my SD card gets Jun 10 10:25:09 SD has a 4bit mode Jun 10 10:25:14 so that would seems to match Jun 10 10:26:05 mickeyl, do you know if the current linux SD code uses the 4bit mode? Jun 10 10:26:15 not offhand Jun 10 10:26:30 RP ? Jun 10 10:27:09 anyone looked @ initng yet ? Jun 10 10:27:21 geentoo are using it now, so it looks mature enough to give it a try Jun 10 10:27:27 I've read the website - not booted it up though Jun 10 10:28:55 mickeyl: well.. nothing can be worse than what gentoo used before I guess Jun 10 10:29:40 mickeyl: (referring to the "mature enough" part) Jun 10 10:32:41 heh, what did they use previously? Jun 10 10:34:54 custom hacks on top of init. basically they're resolving dependencies between init scripts at boot time (which blows up about every 10 boots here forcing me to make use of the reset button...) Jun 10 11:25:20 evening all Jun 10 11:26:10 hey ade|desk Jun 10 11:31:45 hi, i'm crashing on bitbake gpe-image due to the absence of a 'clean' target in the xserver's makefile Jun 10 11:31:49 | make[2]: Entering directory `/home/sisa/oe/build/tmp/work/xserver-xorg-6.8.99.5-r1/xc/include' Jun 10 11:31:49 | make[2]: *** No rule to make target `clean'. Stop. Jun 10 11:33:11 NOTE: package xserver-xorg-6.8.99.5-r1: task do_compile: failed Jun 10 11:40:36 koen|tv, i understand you build for familiar. D Jun 10 11:40:43 Are you crashing with X Jun 10 11:41:03 ? Jun 10 11:41:55 familiar doesn't use xorg Jun 10 11:53:45 It uses's the fonts though Jun 10 11:55:47 could be Jun 10 11:55:51 but not the xserver Jun 10 12:01:09 actually, it doesn't use the fonts either. Jun 10 12:01:29 the xserver-xorg build failure is a bit weird, though. Are you getting any Makefiles generated? Jun 10 12:03:28 pb: we had to patch the fonts to build correctly in the xorg package. Jun 10 12:07:35 I was thinking about the xfbdev the x frame buffer device driver :/ Jun 10 12:11:08 Sony Computer Entertainment, has disclosed in an interview that PlayStation 3 will natively run Linux. In fact, it will come bundled with it, if you purchase the HDD peripheral." Jun 10 12:16:49 hi guys Jun 10 12:16:55 heya spyro! :) Jun 10 12:16:58 mickey|tv: you here? Jun 10 12:17:01 hi ascent Jun 10 12:24:42 what do I type to generate the zImage? Jun 10 12:25:07 hi spyro Jun 10 12:28:23 hi do13 Jun 10 12:31:07 Zero_Chaos: bitbake virtual/kernel Jun 10 12:32:18 koen|tv: thanks Jun 10 12:33:53 * Spyro still cant build tslib. Jun 10 12:34:11 any bb hackers want a look at the debug output ? Jun 10 12:34:41 bitbake shell now has a pastebin command built in Jun 10 12:34:51 koen|tv: whats that then? Jun 10 12:34:57 bitbake -i Jun 10 12:35:04 parse ; build tslib Jun 10 12:35:08 buffers Jun 10 12:35:14 Jun 10 12:35:21 pastebin Jun 10 12:37:51 nice, that is awesome Jun 10 12:39:21 koen|tv: pick right one ? Jun 10 12:39:35 yeah Jun 10 12:39:45 'buffers' should give you a list Jun 10 12:39:54 http://pastebin.com/298532 Jun 10 12:46:16 koen|tv: it appears bitbake is crashing Jun 10 12:46:23 koen|tv: so I cant use this method Jun 10 12:47:14 http://pastebin.ca/13909 Jun 10 12:48:41 Spyro: make sure you have only *one* copy of bitbake around and a recent (say, today) OE tree Jun 10 12:49:14 that line 80 usually indicated an old bitbake Jun 10 12:49:30 which is weird, because the old one doesn't have a shell Jun 10 12:50:25 did you reinstall the setup.py as root after installing the new version of bitbake Jun 10 12:52:57 koen|tv: I'll grab a recent OE Jun 10 12:53:13 jskcr: I dont install my bitbake Jun 10 12:57:55 ibot's dead? Jun 10 13:09:06 koen|tv: where is a recenty OE snapshot? latest I see is 04/may Jun 10 13:09:28 in bk Jun 10 13:09:58 koen|tv: can I get a copy without bk? Jun 10 13:11:01 sure Jun 10 13:11:22 there's a opensource client Jun 10 13:11:32 koen|tv: sure, but its in a shar archive... Jun 10 13:11:34 or wait 10 minutes and wget http://ewi546.ewi.utwente.nl/tmp/OE-20050610.tar.bz2 Jun 10 13:11:39 how TF do you extract that? Jun 10 13:11:49 sh file.shar Jun 10 13:12:02 bah :/ Jun 10 13:12:02 it's a self extracting shell script Jun 10 13:12:18 cant the bk people do ANYTHING the normal way? Jun 10 13:12:38 koen|tv, i remember that famliar doesn't use xserver, but it is building it anyway i think...i did a bitbake gpe-image and it is failing. I consider this a bug as the meta data is provided via oe. Do you think I should notify someone? Jun 10 13:12:40 that would disturbe the force Jun 10 13:12:52 koen|tv: hehe Jun 10 13:12:59 jjg_: it does use an xserver, just not hte one from xorg Jun 10 13:13:05 ok Jun 10 13:13:22 koen|tv, about this bug, what do you recommend? Jun 10 13:13:40 well, it looks like you didn't set a correct MACHINE Jun 10 13:14:02 hm Jun 10 13:14:07 ARCH doesn't do that? Jun 10 13:14:13 no Jun 10 13:14:36 in local.conf? Jun 10 13:14:57 that would work Jun 10 13:15:10 the gettingstarted wiki uses that Jun 10 13:15:22 is that the generally accepted manner? Jun 10 13:15:53 it says so in our gettingstarted document Jun 10 13:16:21 Spyro: tar is almost done Jun 10 13:17:55 i'm looking here and don't see anything...is this the wrong page ? http://oe.handhelds.org/cgi-bin/moin.cgi/GettingStarted Jun 10 13:18:05 that's the one Jun 10 13:18:37 koen|tv: cheers Jun 10 13:19:07 i stupidly removed the comments from local.conf so that i could see the config easier so I am not sure of the syntax, could someone paste in a valid MACHINE directive? Jun 10 13:19:36 what machine do you want to build for? Jun 10 13:19:40 arm Jun 10 13:19:48 what kind of arm? Jun 10 13:20:13 ipaq h5550...arm9? Jun 10 13:20:21 h3900 Jun 10 13:20:32 MACHINE=h3900 ? Jun 10 13:20:36 h5550 is armv5 btw Jun 10 13:20:38 yes Jun 10 13:20:43 good to know Jun 10 13:20:47 thanks koen|tv Jun 10 13:21:00 well, arm9 is armv5 ass well :) Jun 10 13:21:06 ah Jun 10 13:21:08 s/ass/as/ Jun 10 13:21:11 ass Jun 10 13:21:12 ass Jun 10 13:21:16 heh Jun 10 13:21:23 long night Jun 10 13:21:41 i'm trying the build again Jun 10 13:22:12 if it fails, you could try on the the h3900 jffs2 images from http://familiar.handhelds.org/snapshots/v0.8.3/20050610/ Jun 10 13:22:36 so the "image" is really a code repo? Jun 10 13:23:00 no, an image is something you can flash to your device Jun 10 13:23:16 ah, i was really wanting to build from source so that I can build out my own system Jun 10 13:24:07 if you do MACHINE=h3900 and DISTRO=familiar-0.8.3, the results will be pretty much the same Jun 10 13:24:38 so DISTRO=familiar gets latest stable? Jun 10 13:24:39 koen|tv: arm9 is armv4, actually, not armv5. Jun 10 13:25:45 only arm9e is armv5te. Jun 10 13:25:47 jjg_: no, familiar-0.8.2 is the latest stable Jun 10 13:25:53 pb_: aha Jun 10 13:26:07 koen|tv, what happens if DISTRO=familiar? Jun 10 13:26:13 that's what i have now Jun 10 13:26:19 I was just about to edit the arm926 bb again :) Jun 10 13:26:26 koen|tv, btw, looks like adding the MACHINE entry worked Jun 10 13:26:31 jjg_: it will build the latest from everything Jun 10 13:26:58 0.8.3 must be fairly stable if you are recommending it, eh? Jun 10 13:27:42 no, I just want more people testing it Jun 10 13:28:16 :D Jun 10 13:28:41 Spyro: tar is done, wget it if you want Jun 10 13:28:51 i promise to test it, but this particular effort is for a "production R&D" project Jun 10 13:28:59 for Samsung Electronics :) Jun 10 13:29:06 heh Jun 10 13:29:11 yeah, the arm926.conf is correct as it stands. all the arm9x6 parts use arm9e-s cores. Jun 10 13:29:29 in other words, if my demo failed i'd be up the creek without a paddle Jun 10 13:29:32 I guess I should copy prpplague cheat sheet Jun 10 13:30:04 arm920t, which also has a correct .conf file, is an example of an older arm9tdmi part that doesn't have the v5 instructions. Jun 10 13:31:56 thumb, debugger, m...., ICE Jun 10 13:32:06 koen|tv: ? Jun 10 13:32:17 i'm baking my kernel modules now!! :) Jun 10 13:32:37 prpplague: you had a arm cheat sheet next to your screen, right? Jun 10 13:32:39 once oe is running, it actually is pretty great! :) Jun 10 13:33:07 koen|tv: arm asm cheat sheet yea Jun 10 13:33:21 koen|tv: that what you are refering to? Jun 10 13:33:43 ah, I thought it was a modelnumer -> features sheet Jun 10 13:34:11 koen|tv: oh yea, it has that too Jun 10 13:34:23 koen: just copied from that arm developers guide Jun 10 13:36:01 koen: thanks Jun 10 13:39:41 is bitbake smart like make in that if I change some source and rebuild, will it only rebuild the edited file and then link it together? Jun 10 13:41:02 no. Jun 10 13:41:14 it will not redo any tasks unless you tell it to Jun 10 13:42:33 so, for example, if i bitbake gpe-image ...then go to the source directory, change a file and rerun "bitbake gpe-image" ... how does oe behave ? Jun 10 13:43:08 it does nothing. Jun 10 13:43:14 hm Jun 10 13:43:31 as far as it's concerned the task has completed Jun 10 13:43:32 how would i tell it to rebuild that file and relink everythign? Jun 10 13:43:41 run clean on that task Jun 10 13:43:43 and then rebuild Jun 10 13:43:57 what exactly is a "task" in oe world? Jun 10 13:44:11 the bitbake command? Jun 10 13:44:18 eg., "bitbake gpe-image" Jun 10 13:44:24 some thing like do_build or do_package for a bb Jun 10 13:44:29 ah Jun 10 13:44:31 one phase of a package build: fetch, patch, configure, compile, and so on Jun 10 13:44:33 in the bb file? Jun 10 13:44:36 right Jun 10 13:44:38 gotcha Jun 10 13:44:50 in the case of changing source, bb -b foo.bb -c clean Jun 10 13:44:55 will clean and tell it to start over Jun 10 13:44:59 so the task is for that particular package Jun 10 13:45:14 yes Jun 10 13:45:45 bitbake, at least in its current form, isn't really ideal for interactive debugging. Jun 10 13:45:47 so i need to manually rerun the task for the particular package i edit, then what would be the command to link it all together again? Jun 10 13:45:58 pb_, i see Jun 10 13:46:05 you just rerun the build task Jun 10 13:46:28 and since i manually rebuilt the package...all obj files are available and it just relinks Jun 10 13:46:42 right Jun 10 13:46:47 ok Jun 10 13:47:27 are devs using something like crosstool to do interactive debugging...then just merge their final changes into the oe code tree? Jun 10 13:48:06 oe doesnt store individual source Jun 10 13:48:15 I do 99% of my interactive debugging natively. I'm not quite sure what the qt/e crowd do for that. Jun 10 13:48:28 really? doesn't that take forever? Jun 10 13:48:50 pb_: Mostly development on whatever desktop platform they use Jun 10 13:49:07 the native stuff? no, not really. I have a 2GHz athlon or something. Jun 10 13:49:12 treke, you use oe for your day to day dev? Jun 10 13:49:22 jjg_: no Jun 10 13:49:37 it isn't an appropriate tool for that Jun 10 13:50:29 treke, ok, good to know. what do you recommend? Jun 10 13:50:38 I just do native development Jun 10 13:50:39 like pb Jun 10 13:51:39 for the 1% of debugging that's target specific, I either use oe with frequent clean/rebuild cycles, or (more often) just add tmp/staging/... to my $PATH and cross compile stuff by hand. Jun 10 13:52:05 in my case, almost nothing is target specific Jun 10 13:52:57 right Jun 10 13:53:24 other than the typical "your code runs find on a 2.6ghz xeon, but not so well on a 206mhz strongarm" issue Jun 10 13:59:11 03mickeyl * r249 10bitbake/lib/bb/shell.py: Jun 10 13:59:11 OO overhaul: Jun 10 13:59:11 - Introduce classes for commands and the command processor Jun 10 13:59:11 - Autoregister available commands Jun 10 13:59:11 - Fix 'shell' command to work with unlimited number of parameters Jun 10 13:59:12 - Remove 'exec' command Jun 10 13:59:13 * koen considers installing distcc on his batch of ARMs Jun 10 13:59:30 koen *g* Jun 10 14:00:19 running OE inside scratchbox using distcc Jun 10 14:00:22 hmmmmm Jun 10 14:00:57 I should be able to put in some vmware or colinux to complete the foolishness Jun 10 14:21:27 * chouimat|pub is away: PUB Jun 10 14:41:06 * prpplague heads home for the weekend Jun 10 14:51:55 03pb 07 * r1.3538 10openembedded/packages/xserver/xserver-xorg_6.8.99.5.bb: fix up imake slightly for cross compiling Jun 10 14:52:25 * treke- still has nightmares of using imake Jun 10 14:53:11 heh, yeah Jun 10 14:53:26 I think that should fix the "no rule to make clean" thing that people have been seeing with xserver-xorg, anyway. Jun 10 14:53:27 of course that was a long time ago Jun 10 14:53:33 and I was much dumber then Jun 10 14:53:38 it hasn't gotten any better since then Jun 10 14:53:41 hehe Jun 10 14:53:46 back in the 3.9 days :p Jun 10 17:39:04 anyone know how to resolve the "no rule to make clean" problem with xserver-xorg ? Jun 10 19:26:34 * chouimat|pub is back. Jun 10 19:55:32 hmm Jun 10 19:55:43 tmp/work uses lots of space if not emptied ;) Jun 10 20:05:43 Has there been any consideration into splitting openembedded? eg OE base (stuff that applies to all targets), OE opie, OE gpe, etc? Jun 10 20:13:35 busybox doesnt like me Jun 10 20:15:21 ljp: how so? Jun 10 20:16:51 try to login (first boot as root) 'login: cannot run /bin/sh: No such file or directory Jun 10 20:21:06 hmm... :( Jun 10 20:40:13 weird Jun 10 20:53:06 | /bin/sh: /home/luke-jr/src/oe2005/build.alt/tmp/cross/arm-linux/bin/ccache: No such file or directory Jun 10 20:53:11 What would provide a cross-ccache? Jun 10 20:53:25 or is the problem that xorg-x11 wants to use it? Jun 10 20:54:37 err... xserver-xorg Jun 10 23:41:09 hmm.. anyone have knights.tar.gz ? Jun 10 23:49:22 awol Jun 10 23:57:24 ;( **** ENDING LOGGING AT Fri Jun 10 23:59:57 2005