**** BEGIN LOGGING AT Fri Apr 13 02:59:56 2007 Apr 13 04:01:46 Anyone have a link to this file? sqlite-3.3.9.tar.gz Apr 13 04:05:46 Zero_Chaos: http://sources.nslu2-linux.org/sources/sqlite-3.3.9.tar.gz Apr 13 04:06:15 rwhitby: thanks man, do you have commit access to .oz354x? or should I file a bug? Apr 13 04:06:34 I don't deal with that branch, no. Apr 13 04:07:26 rwhitby: thanks again for the link. Apr 13 04:09:07 Continued download failed on this file, which conflicts with `-c'. Apr 13 04:09:08 Refusing to truncate existing file `/home/zero/oe/sources/sqlite-3.3.9.tar.gz'. Apr 13 04:09:13 complaining the file is too big? Apr 13 04:18:11 !oebug 2106 Apr 13 04:18:13 * * Bug 2106, Status: NEW, Created: 2007-04-12 21:17 Apr 13 04:18:13 * * r.farina(AT)adelphia.net: sqlite 3.3.9 fetch failure Apr 13 04:18:14 * * http://bugs.openembedded.org/show_bug.cgi?id=2106 Apr 13 04:19:05 if anyone can resolve that fetch issue, that would be great. Apr 13 05:11:43 Hi! Apr 13 05:21:09 what is the status of http://www.openembedded.org/wiki/PackagedStaging Apr 13 05:22:31 I have bitbaked ipkg-native, and would like to create a image on my host platform Apr 13 05:22:53 I guess that happens for every distro but i don't know how :p Apr 13 06:55:06 hrw|gone: when will you be checking in the task-base rewrite? I have some usb wifi dongle updates to merge into it. Apr 13 06:59:19 morning Apr 13 06:59:24 rwhitby: after test Apr 13 07:00:00 Do you have a latest version I can download and use locally to test? Apr 13 07:00:20 I mailed latest one Apr 13 07:02:01 good morning all Apr 13 07:02:14 morning koen Apr 13 07:02:36 hey Ifaistos Apr 13 07:02:41 hrw: can you perhaps check it into the contrib area or something, so we can symlink to it and test the latest as you update it? Apr 13 07:04:31 hrw: I suspect the distro-ssh-daemon can't be set to "" Apr 13 07:04:52 hrw: FOO = "", FOO ?= "bar" -> FOO == bar Apr 13 07:06:14 ~lart libdaemon Apr 13 07:06:14 * ibot shoves a crumpet down libdaemon's throat, happy now?! Huh? Want some JAM with that? Apr 13 07:06:36 rwhitby: I will do test today and compare result with older image. Apr 13 07:06:43 if it will be ok then I will push it Apr 13 07:06:56 hrw: I'm just about to flash a mokoslug image built from it. Apr 13 07:08:04 Isn't diffing the manifest a better way? Apr 13 07:08:23 meh - real men just flash Apr 13 07:08:52 "it boots" is not a garantee for everything being in place Apr 13 07:09:14 of course diffing the manifest is the better way. Apr 13 07:09:46 I diff ipkg status file Apr 13 07:10:05 diffing the manifest is what you do while you're waiting for it to flash and boot ... Apr 13 07:11:04 bbiab Apr 13 07:12:08 likewise: good morning! Apr 13 07:12:12 "XScale DSP coprocessor detected. Apr 13 07:12:13 " - hmm, hadn't noticed that before. must be something new in the kernel. Apr 13 07:12:33 good morning! Apr 13 07:13:13 rwhitby: http://lists.arm.linux.org.uk/pipermail/linux-arm-kernel/2006-October/036330.html Apr 13 07:14:19 likewise: thx for the pointer Apr 13 07:14:45 rwhitby: not sure if this is the exact patch that introduced this though. Apr 13 07:15:03 rwhitby: at least it's a lennert patch, so it must be good Apr 13 07:18:21 Hmm: /home/slug/mokoslug/tmp/rootfs/usr/lib/ipkg/info/udev.postinst: line 10: /etc/udev/mount.blacklist: Permission denied Apr 13 07:20:28 03hrw 07org.oe.dev * rcb19c88a... 10/ (5 files in 2 dirs): dates: update to 0.4.1 (from Poky) Apr 13 07:20:38 03hrw 07org.oe.dev * r5bcded2d... 10/ (3 files in 3 dirs): tasks: Patch tasks to make it single instance (from Poky) Apr 13 07:23:00 hrw: 'nfs' is listed twice in PACKAGES (once for machine features and once for distro features), but is then only added to task-base depending on distro features. Why is that? Apr 13 07:24:18 it should not be in machine part.. Apr 13 07:25:55 koen: if nfs is a distro feature, and is enabled in angstrom, and I don't want nfs included in the mokoslug image (cause it's too big and you don't need it to boot a mokoslug - you can ipkg install it later), how can I migrate mokoslug from a distro to an image? Apr 13 07:26:29 rwhitby: mokoslug image == task-boot + task-base-bluetooth + task-... Apr 13 07:27:48 hrw: this is not a task-base related question Apr 13 07:27:56 ok Apr 13 07:28:43 what in OE use fuser? Apr 13 07:29:28 nothing I know off, but fuser and setserial were added because someone added them to BOOTSTRAP* at some point in time Apr 13 07:29:46 IMO fuser and setserial don't belong in task-base Apr 13 07:30:00 fuser is in busybox 1.2.1 Apr 13 07:30:10 I'm seeing whether it really is possible to make mokoslug just another type of angstrom image, or whether it is forced to be a distro by these types of things. Apr 13 07:31:19 koen: task-base-serial? with lszrz and setserial Apr 13 07:31:41 hrw: that's an option Apr 13 07:32:03 morning all Apr 13 07:32:09 hi RP Apr 13 07:33:23 hrw: yep, nslu2 would not have task-base-serial by default, so that's good for us. Apr 13 07:33:56 rwhitby: I want to generate some task-base-* nevermind which features are set in distro/machine Apr 13 07:34:33 in my opinion, you should generate them all, cause it makes it easy to install stuff later Apr 13 07:34:45 (generate, but not include in task-base) Apr 13 07:35:04 (not be a run-time dependency of task-base I should say) Apr 13 07:35:12 good idea, but some of them are useless on certain machines Apr 13 07:35:30 I think that's outside the scope of task-base and would fit better in task-generic (or something like that) Apr 13 07:35:34 task-base-acpi on !x86*, task-base-usbhost on machines which lack usbhost Apr 13 07:36:22 hrw: what if someone with a machine which doesn't have usbhost by default, but they put in a minipci card with provides usb host ports ... Apr 13 07:36:33 they'd like to be able to ipkg install task-base-usbhost .. Apr 13 07:36:52 (this is the situation for nslu2 and serial, for instance) Apr 13 07:37:03 http://freshome.com/2007/04/12/screens-of-3-and-9-millimeters-from-sony/ Apr 13 07:37:16 [09:35] koen: I think that's outside the scope of task-base and would fit better in task-generic (or something like that) Apr 13 07:38:04 not necessary koen Apr 13 07:38:18 rwhitby: If you have problems like nfs, please do keep a list of them. We'll then have to see if we can resolve them somehow... Apr 13 07:38:31 koen: task-base.bb can generate task-base-* but only part of them will be used in task-base Apr 13 07:38:43 hrw: yes, that's what I tried to say. Apr 13 07:39:09 especially when we do one change: Apr 13 07:39:10 PACKAGE_ARCH = "${MACHINE_ARCH}" Apr 13 07:39:22 hrw: I know it can, but if it isn't machine or distro specific, why put it in task-base? Apr 13 07:39:23 -> PACKAGE_ARCH_task-base = "${MACHINE_ARCH}" Apr 13 07:39:49 I would have thought serial was machine specific ... Apr 13 07:40:16 rwhitby: A task-serial task package isn't machine specific Apr 13 07:40:30 rwhitby: Whether its included by default is though Apr 13 07:40:33 so I don't want task-base-serial included in the nslu2 rootfs, but I do want it available to ipkg install. Apr 13 07:40:38 most of task-base stuff is not machine/distro related Apr 13 07:40:40 (machine specific = ${@base_contains("MACHINE_FEATURES",......) Apr 13 07:40:44 RP: agreed. Apr 13 07:42:06 so what I am saying is that just because the currently defined machine or distro features do not enable a certain task-base-* package, please generate the package anyway (but do not make it a dependency of task-base) so that people who "upgrade" the features of their machine can ipkg install those packages later. Apr 13 07:42:35 rwhitby: hrw's proposal does that Apr 13 07:42:59 RP: agreed. it sounded like koen was arguing against that. Apr 13 07:43:18 "[09:35] koen: I think that's outside the scope of task-base and would fit better in task-generic (or something like that)" Apr 13 07:43:32 but I may have misunderstood what koen said ... Apr 13 07:44:08 so I'm trying to state it unambiguously, so it can be either supported or opposed cleanly. Apr 13 07:44:13 I'm not against it, I just think tasks without ${@base_contains(... should be outside task-base Apr 13 07:44:39 koen: we're talking at crossed purposes. Apr 13 07:44:54 you're talking about tasks that are independent of machine or distro features, right? Apr 13 07:45:03 yes Apr 13 07:45:12 PACKAGE_ARCH = "all" Apr 13 07:45:37 I'm talking about tasks that are dependent on either machine or distro features, but are not enable for the current configuration - I want those packages generated even though they are not selected to be dependencies of task-base. Apr 13 07:45:54 if we're going to overhaul task-base, we might as well clean it up Apr 13 07:46:09 koen: do you agree or disagree with my last statement? Apr 13 07:46:33 I agree that a task without a base_contains shouldn't be in task-base. Apr 13 07:47:53 * koen is trying to see if users can shoot themselves in the foot with such packages Apr 13 07:48:31 Did anyone test the dependency issues patch I posted? Is it too soon to commit that yet? :) Apr 13 07:48:41 the packages looks pretty safe Apr 13 07:48:45 another example. if I add a usbvga adapter to a slug, then I might want to ipkg install packages that are dependent on the screen machine feature ... Apr 13 07:49:00 rwhitby: I think it can't hurt to generate such packages Apr 13 07:49:11 rwhitby: I think we're agreed that there should be the task packages available Apr 13 07:49:12 I am thinking about task-base also as 'something which provide base support for features' not only as 'bootstrap vars replacement' Apr 13 07:49:19 morning Apr 13 07:49:24 hi XorA Apr 13 07:49:24 hi XorA Apr 13 07:49:28 hey XorA Apr 13 07:49:47 koen, RP, hrw: sounds like we're all in agreement then. Apr 13 07:51:25 http://pastebin.ca/437768 - current version Apr 13 07:52:50 hrw: could you do DISTRO_SSH_DAEMON ="" in local.conf and check the output? Apr 13 07:52:57 ok Apr 13 07:53:03 last time I tried such a construction didn't work Apr 13 07:54:05 Depends: task-boot, task-distro-base, task-machine-base, task-base-kernel26, task-base-acpi,.... Apr 13 07:54:35 with DISTRO_SSH_DAEMON ="" removed I got dropbear again Apr 13 07:55:07 ok, thanks for checking Apr 13 07:55:32 ~lart pastebin.ca putting ^M's in the raw download ... Apr 13 07:55:33 * ibot forces pastebin.ca putting ^M's in the raw download ... to use Outlook Express Apr 13 07:55:55 heh - they would come out fine in Outlook Express. no botsnack. Apr 13 07:57:42 ~ugt Apr 13 07:57:48 i guess ugt is Universial Greeting Time. Created in #mipslinux, it is a rule that states that whenever somebody enters an IRC channel it is always morning, and it is always late when the person leaves. The local time of any other people in the channel, including the greeter, is irrelevant. http://www.total-knowledge.com/~ilya/mips/ugt.html Apr 13 07:58:17 hrw: how about ${@base_contains("COMBINED_FEATURES", "pci", "madwifi-ng-modules", "",d)} in task-base-wifi, for the PCI atheros cards in nas100d and dsmg600 Apr 13 07:58:33 (and similarly for madwifi-ng-tools) Apr 13 07:59:05 and ${@base_contains("COMBINED_FEATURES", "usbhost", "kernel-module-zd1211rw", "",d)} for task-base-wifi for zd1211 usb wifi dongles for nslu2 Apr 13 07:59:21 added Apr 13 08:00:15 is "bridge-utils" in the same league as "wireless-tools" and "wpa-supplicant" for task-base-wifi? Apr 13 08:00:33 rather not Apr 13 08:00:47 wireless-tools are used by most of wifi drivers Apr 13 08:00:50 * rwhitby is looking through task-mokogateway.bb to see what should be in base and what should not. Apr 13 08:00:56 wpa-supplicant is needed for WPA Apr 13 08:01:50 hrw: it might be useful to define a criteria for what should be in an what should be out, and put that as a comment in the file for each task-base-* package. Apr 13 08:02:22 e.g. for wifi, it should be only what is required to automatically bring a wifi interface up and enable you to ssh in via that interface so you can run ipkg install to get anything else you need. Apr 13 08:02:32 sure Apr 13 08:02:38 (that criteria would exclude bridge-utils, as expected) Apr 13 08:10:03 anyone object to a task-base-ata (or pata or sata) which has those things required to boot a rootfs from an attached disk (e.g. nas100d, dsmg600, ...) Apr 13 08:11:39 task-base-ata sounds best Apr 13 08:12:46 when my builds end I will push task-base.bb Apr 13 08:15:23 yay I now have USB OTG adapter that is not a cable Apr 13 08:21:02 re: hostap-utils. That's a prism-specific package, right? Apr 13 08:21:20 yes Apr 13 08:22:07 how do we allow it to be excluded for devices that have a non-prism wifi? Apr 13 08:22:20 hostap support pci/pcmcia/cf cards Apr 13 08:22:34 rwhitby: minipci feature? Apr 13 08:22:55 Atheros minipci don't use hostap-utils, does it? Apr 13 08:23:07 does not Apr 13 08:23:20 iirc there is no minipci cards with prism Apr 13 08:23:49 ah, sorry, I didn't follow you at first. Apr 13 08:23:57 that would solve the immediate problem Apr 13 08:24:16 what about the more general problem of packages that are in task-base, but only support one type of hardware? Apr 13 08:24:31 for kernel modules it's ok, cause you don't build the kernel module if your machine doesn't support it. Apr 13 08:24:51 but for packages like hostap-utils and madwifi-ng-tools, we need a more general solution. Apr 13 08:26:02 wifi-prism wifi-orinoco wifi-acx100 wifi-atheros wifi-zd1211 wifi-ralink wifi-wlanng wifi-bcm43xx? Apr 13 08:26:32 I dunno the solution, just putting the problem on the table :-) Apr 13 08:26:45 looking at how openwrt do it might help. Apr 13 08:29:16 rwhitby: added MACHINE and DISTRO_FEATURE = "redboot" on the slug yet? :) Apr 13 08:29:39 NAiL: not yet. Apr 13 08:29:49 It's not really useful, actually Apr 13 08:31:01 redboot doesn't read fconfig on the slug, does it? Apr 13 08:31:29 hrw: openwrt do: "bcm43xx,ipw2100,ipw2200,airo,hermes,hermes-pci,hermes-plx,prism54,zd1211rw" Apr 13 08:32:08 hmm - I will need to get them to add atheros for the nas100d ... Apr 13 08:32:25 ok. angstrom/openmoko-image built Apr 13 08:32:26 Is anyone actually using hermes-plx? Apr 13 08:32:46 NAiL: which format it is? Apr 13 08:33:12 bbiab Apr 13 08:33:17 IIRC, it's pcmcia in a pci-slot. The plx is a special converter-thingy. Apr 13 08:33:31 ah Apr 13 08:34:28 I used it when I worked at Siemens, to get their wifi cards working. Piece of crap ;) Apr 13 08:39:31 progear/angstrom built too Apr 13 08:40:04 task-base.bb | 391 +++++++++++++++++++++++++++++++++-------------------------- Apr 13 08:40:04 1 file changed, 219 insertions(+), 172 deletions(-) Apr 13 08:40:15 nice.. Apr 13 08:44:28 * hrw pushed task-base Apr 13 08:45:34 03hrw 07org.oe.dev * r920ac556... 10/ (1 packages/tasks/task-base.bb): (log message trimmed) Apr 13 08:45:34 task-base: major restructure Apr 13 08:45:34 - task-base-FEATURENAME are built nevermind does MACHINE supports them Apr 13 08:45:34 - task-base depends on task-* packages only - MACHINE/DISTRO features only Apr 13 08:45:34 - task-boot contain stuff needed to get minimal bootable image Apr 13 08:45:35 - task-distro-base contain distro addons Apr 13 08:45:37 - task-machine-base contain machine specific addons Apr 13 08:46:30 philippe: osb-browser and gpe-mini-browser work great on armv5te/EABI Apr 13 08:54:27 rwhitby: we do not build our own redboot for slug, do we? Apr 13 08:54:49 rwhitby: as for our custom targets we do, just not from within OE (yet). Apr 13 09:17:20 rwhitby: When you say task-base-ata, should it not be task-base-pivotroot by the description you gave? Apr 13 09:26:17 likewise: no, we don't build redboot. what custom targets are you referring to? Apr 13 09:26:30 rwhitby: proprietary. Apr 13 09:26:39 RP: no, this is just the tools required to get the disk accessible. altboot would do the rest. Apr 13 09:26:49 rwhitby: Just wanted to know if there was effort to build redboot from within OE, I could help. Apr 13 09:27:13 likewise: we don't reflash the first stage bootloader, and redboot is too big for a second stage :-) Apr 13 09:28:01 rwhitby: Yes, got the policy :-) Apr 13 09:32:19 * rwhitby wonders about MACHINE_TASK_PROVIDER and every man and his dog forcibly defines it to it's default value ... Apr 13 09:35:48 rwhitby: I wonder about MACHINE_FEATURE_pcmcia_RRECOMMENDS ;D Apr 13 09:37:41 hrw: one problem with building all task-base packages unconditionally - some of them don't build for me :-( Apr 13 09:38:58 rwhitby: which one? Apr 13 09:39:50 hmm - scratch that, it was just a corrupted download of acpid Apr 13 09:44:52 hrw: where is MACHINE_FEATURE_pcmcia_RRECOMMENDS defined or used? Apr 13 09:45:03 * rwhitby doesn't get the reference ... Apr 13 09:45:21 rwhitby: in machine config Apr 13 09:45:42 MACHINE_FEATURE_pcmcia_RRECOMMENDS = "kernel-module-strange-pcmcia-controller-driver" for example Apr 13 09:46:37 * rwhitby looks at MACHINE_FEATURE_pcmcia_RRECOMMENDS but still can't parse what it means. Apr 13 09:47:38 ok Apr 13 09:48:03 its extra packages needed for machine to support 'pcmcia' feature Apr 13 09:50:42 ah, so task-base would use it. I thought it was some new funky override syntax that I hadn't seen before ... Apr 13 09:53:29 ok, I have collie booted Apr 13 09:59:18 hi mickeyl Apr 13 09:59:33 yo mickeyl Apr 13 09:59:41 morning guys! Apr 13 09:59:45 how is everyone doing today? Apr 13 09:59:55 collie screen SUXX Apr 13 09:59:59 hehe Apr 13 10:00:02 you've been spoiled Apr 13 10:00:04 I forgot how bad it is Apr 13 10:00:18 but 2.6 is working ;) Apr 13 10:00:40 coolness! Apr 13 10:01:26 * XorA booted 2.6 but zero luck with SD cards Apr 13 10:01:45 hi mickeyl Apr 13 10:02:41 hi richard Apr 13 10:04:35 now.. where I have 16mmc Apr 13 10:06:42 16MB mmc works Apr 13 10:09:50 hi all Apr 13 10:10:54 Does somebody already try OE with an IPaq HX2790? Apr 13 10:11:36 morning mickeyl: how are *you* doing today? Apr 13 10:12:46 likewise: not bad. despite the usual stress that comes when you are working on three "projects" at the same time, i'm feeling pretty well :) Apr 13 10:15:03 any plans for next OEDEM yet? Apr 13 10:15:34 we need one Apr 13 10:15:40 there's so many things to discuss Apr 13 10:15:55 but i'm afraid i can't participate before July Apr 13 10:16:14 so somewhere July/August would work for me best Apr 13 10:16:19 mickeyl: probably same timescale for me, Id have to save up Apr 13 10:16:27 hrw: we still plan to do it yat your place? Apr 13 10:16:58 XorA: well, this time i want to have some kind of sponsoring Apr 13 10:18:22 mickeyl : hrw's place. Would that be Poland? And where? Apr 13 10:19:13 In July/August timeframe I am going back to .be so I will be in .de and taking the boat in Rostock which is not to far from the .pl border. Would be nice if I could join. Apr 13 10:20:20 hey mickeyl Apr 13 10:21:31 philippe: Poznan, Poland Apr 13 10:22:15 philippe: did you get my message about osb-browser and gpe-mini-browser? Apr 13 10:22:56 philippe: yeah, that'd be cool Apr 13 10:22:58 koen: nope sorry. Apr 13 10:23:01 morning koen Apr 13 10:23:05 hrw: hmm, seems doable Apr 13 10:23:17 philippe: both of them work on my hx4700 Apr 13 10:23:23 (svn versions) Apr 13 10:23:29 koen: YAY!!! Apr 13 10:23:55 And some more fixes just have gone in today. I think they solved a double free issue I have been seeing. Apr 13 10:24:02 * philippe is building atm to verify Apr 13 10:26:51 the new icons for pimlico are awesome: http://www.andreasn.se/blog/?p=47 Apr 13 10:27:53 koen: pimlico website use them Apr 13 10:28:05 I noticed Apr 13 10:28:21 hrw: does poky have an 'n800' machine? Apr 13 10:28:26 yes Apr 13 10:28:28 nokia800 Apr 13 10:28:31 ok Apr 13 10:28:40 I built image for it today Apr 13 10:29:45 I'm only using my 770 at the moment, so I thought I could use the n800 for testing Apr 13 10:30:06 hrw: my card results pretty much match yours Apr 13 10:30:24 hrw: except microSD says error switching card to high speed mode Apr 13 10:30:35 hrw: but Ill get exact errors over weekend and do proper email Apr 13 10:39:09 philippe: webcore does seem to have bugs with 200dpi rendering: http://scap.linuxtogo.org/ Apr 13 10:42:12 koen: it is in feeds? Apr 13 10:42:52 yes, but the libpackaging sucks, so you have to use -force-overwrite to get around problems with images Apr 13 10:43:35 (if you have webcore already installed) Apr 13 10:45:10 NOTE: package osb-jscore-0.5.0+cvs20060212: started Apr 13 10:45:18 a bit old... Apr 13 10:45:42 koen: angstrom use this ver or newer? Apr 13 10:45:51 it uses svn Apr 13 10:46:08 I removed all !svn recipes for my webcore dir Apr 13 10:46:14 hmmm Apr 13 10:46:21 rm_work doesn't rm with bitbake 1.8 Apr 13 10:46:36 koen: then revert them and add proper PREF_VER into angstrom... Apr 13 10:49:51 koen: in whichsense? Unless you mean the somewhat flacky displaying of the OE logo. Apr 13 10:50:05 philippe: check the google page Apr 13 10:50:17 everything is too big Apr 13 10:50:18 hrw: current gtk-webcore svn is a lot better than that version. Apr 13 10:50:44 koen: aha, you mean the size of the entry box? Apr 13 10:50:51 yes Apr 13 10:51:00 and the OE page seems to be too wide as well Apr 13 10:51:53 Hmmm, I will look into it. See if I can find anything. HAve you tried with gpe-mini-browser? It could be that flower uses a too big font to start with Apr 13 10:52:19 the same Apr 13 10:53:09 koen: ok. I'll see what I can do. Apr 13 10:53:55 philippe: http://code.google.com/soc/openmoko/appinfo.html?csaid=B5A2E96741FD60E4 Apr 13 10:57:14 03koen 07org.oe.dev * re1b0cb5c... 10/ (1 conf/distro/angstrom-2007.1.conf): angstrom: add versions for gtk-webcore Apr 13 11:00:15 koen: sounds very good :-) Let's hope they succeed. Apr 13 11:03:52 mickeyl a mentor for zecke ... sounds fun Apr 13 11:04:16 unfortunately mickeyl is no longer a student, then they could reverse their roles Apr 13 11:04:49 somehow I have the feeling that in neither case anyone would need much mentoring :-) Apr 13 11:06:21 schurig: i sincerely hope so, since i have not much time for mentoring anyway ;) Apr 13 11:06:59 * koen wishes someone would get hired to finish packaged-staging Apr 13 11:07:30 bitbake 1.8 would make that a lot cleaner Apr 13 11:08:43 koen: what parts are missing? Apr 13 11:09:24 schurig: proper locking, package format independance and conflict resolution Apr 13 11:19:25 philippe: http://scap.linuxtogo.org/files/5293992779243dfba377172d420ccf64.png <- notice how it always seems to think my screen is ~25 percent wider Apr 13 11:19:45 philippe: I suspect something goes wrong with the windowsize calculation Apr 13 11:22:59 koen: do you have some time for me? Apr 13 11:24:49 Does somebody already try OE with an IPaq HX2790? Apr 13 11:25:32 hrw: could you check for packages using task-base-core-default, before removing it? Apr 13 11:26:13 ops Apr 13 11:28:23 03koen 07org.oe.dev * rea2cdd5b... 10/ (7 files in 2 dirs): angstrom: remove references to task-base-core-default Apr 13 11:30:00 (OT) Is anyone aware of size limitations of an initramfs? Apr 13 11:30:20 likewise: isn't it the same as the max ramdisk size? Apr 13 11:30:37 iirc 4MB by default, Kconfig option Apr 13 11:31:09 koen: I thought the two were unrelated, but now you mention 4MB default... :-) Apr 13 11:32:01 does anybody already configure pxafb on Ipaq hx2790? I don't how to find my pixclock? sorry for this newbie question Apr 13 11:33:32 manu___: tried #handhelds.org? Apr 13 11:34:05 hrw: hx2xxx is RP-ware Apr 13 11:34:10 k thx Apr 13 11:34:27 koen: ah.. yes.. Apr 13 11:35:03 manu___: use OE, MACHINE=hx2xxx DISTRO=angstrom-2007.1 bitbake angstrom-console-image Apr 13 11:35:39 k thx i will try Apr 13 11:48:58 zecke: moin Apr 13 11:49:08 hija Apr 13 11:52:46 zecke: gtk-webcore svn works on ARM again Apr 13 11:53:13 koen: what was the arm related issue? Apr 13 11:53:22 segfaults all over the place Apr 13 11:56:18 hello everybody Apr 13 11:56:27 <4>Alignment trap: gpe-mini-browse (1257) PC=0x4041ced4 Instr=0xe893000c Address=0x4045b226 FSR 0x013 Apr 13 11:56:32 thesing: hi Apr 13 11:56:41 hey thesing Apr 13 11:57:03 hrw: So you have one working card and two nonworking. Apr 13 11:57:51 thesing: and two not detected at all (rs-mmc and microsd) Apr 13 11:58:29 hrw: can you build the mmc modules with debug support, replace them and try again? Apr 13 11:58:43 probably Apr 13 11:58:45 hrw: i can upload the recompiled modules if you like. Apr 13 11:59:59 hrw: you could also add some extra debug info by doing : echo "1" > /sys/bus/locomo-bus/drivers/locomo-spi/verbose Apr 13 12:00:59 hrw: it's a pity, that all my card work. Apr 13 12:01:17 thesing: you live in Berlin? Apr 13 12:02:35 about 100 m afar from berlin. Apr 13 12:03:39 thesing: mail me your address - I will send you some card or cards Apr 13 12:03:56 that would be great. Apr 13 12:04:43 thesing: I also tried mmc driver and I got pretty much indentical results to hrw so I shall wait until you manage to get those cards in your hands and try again Apr 13 12:05:34 * hrw checks 256M SD Apr 13 12:06:51 same ;( Apr 13 12:07:07 thesing: so mail me address and I will send you 64 and 256M cards Apr 13 12:07:20 I will do. Apr 13 12:07:29 thesing: you have other arm mmc machine to compare results? Apr 13 12:07:30 hrw: collie? Apr 13 12:07:35 Laibsch: yes Apr 13 12:07:49 nice. So you are making progress? Apr 13 12:08:09 hrw: I guess you know but my problems with nunome were coming from stupid dash Apr 13 12:08:19 Laibsch: I saw that Apr 13 12:09:08 hrw: yes. I own a borzoi, which my girlfriend has occupied. But with a little fight, I'll get it back ;) Apr 13 12:09:43 Oh, Thomas Kunze himself! Apr 13 12:09:58 ~hail Thomas Kunze, the saviour of collie ;-) Apr 13 12:10:19 * ibot bows down to Thomas Kunze, the saviour of collie ;-) and chants, "I'M NOT WORTHY!!" Apr 13 12:10:52 thesing: Your work on the driver is very much appreciated. At some point in time you should consider claiming the bounty that is out there. Apr 13 12:11:40 \me has to get some food. Apr 13 12:13:52 hrw: http://rafb.net/p/87NCG226.html Apr 13 12:14:14 looks pretty ok Apr 13 12:14:48 great Apr 13 12:16:58 heh Apr 13 12:17:07 I did found a bug in OE :) Apr 13 12:17:10 find* Apr 13 12:17:19 koen@bitbake:/data/build/koen/OE/build/tmp/angstrom$ find deploy/glibc/ipk/ -name task-base_* | grep r26 Apr 13 12:17:19 deploy/glibc/ipk/all/task-base_1.0-r26_a780.ipk Apr 13 12:17:19 deploy/glibc/ipk/all/task-base_1.0-r26_h6300.ipk Apr 13 12:17:43 koen: its bug in your change to package.bbclass Apr 13 12:18:04 that's another way of putting it Apr 13 12:19:13 arch = bb.data.getVar('PACKAGE_ARCH', d, 1) Apr 13 12:19:13 outdir = "%s/%s" % (outdir, arch) Apr 13 12:19:29 that seems to get confused by subpackages with another arch Apr 13 12:19:40 hmm, forms dont seem to work in gpe-mini-browser, I cant search on google Apr 13 12:20:08 * thesing is back. Apr 13 12:21:25 thesing: Nice work on the driver. We should look at getting that merged into OE :) Apr 13 12:21:53 thesing: I mean into linux-rp so the support is in the standard zaurus kernels :) Apr 13 12:21:58 RP: I already posted a mtn diff at oz-devel. Apr 13 12:22:09 RP: I'm testing it now and will push Apr 13 12:22:20 thesing: Right, I'm just lacking time atm :-( Apr 13 12:22:25 FATAL: This kernel is too big for your Collie and will not be flashed. Please reduce the size of the kernel by making more of i Apr 13 12:22:36 when I will pass this I will push Apr 13 12:22:44 XorA: weird. Seems to work fine here. Just tried google and it works for me. Apr 13 12:23:04 hrw: gcc3 or gcc4? Apr 13 12:23:21 and amd64? Apr 13 12:23:43 I used gcc4 and amd64 and it worked for me. Apr 13 12:23:45 gcc3 amd64 Apr 13 12:23:57 philippe: if I goto www.google.co.uk it works, if I goto www.google.com it doesnt Apr 13 12:23:57 hrw: right, that's causing your bug Apr 13 12:24:15 gcc3 on amd64 == bigger binaries Apr 13 12:24:30 and I lack oabi gcc4 toolchain Apr 13 12:24:57 philippe: also www.google.com (which does a redirect to .co.uk) the image doesnt load Apr 13 12:25:07 XorA: google.com redirects me to google.fi automatically... Apr 13 12:25:08 philippe: so I guess redirects are slightly iffy Apr 13 12:25:23 I'll push Apr 13 12:25:46 XorA: could be. I haven't been able to test it on ARM yet. It works fine on my desktop though Apr 13 12:26:01 koen: Supposedly the linux kernel decompressor (at least for ARM) has a 4MB total limitation, that leaves ~3MB for initramfs. At least that's something I remember but cannot find the claim anymore. Apr 13 12:26:18 koen: New logo proposal for OE: http://tango.freedesktop.org/static/cvs/tango-icon-theme/scalable/emotes/face-monkey.svg Apr 13 12:26:21 philippe: although it is now working as I would expect, maybe our stupid proxy is serving duffness Apr 13 12:26:35 like2wise: :) Apr 13 12:26:36 philippe: I will try again tonight without a braindead squid in the way Apr 13 12:26:45 Build Your Own Build Monkey Apr 13 12:27:12 RP: could you have a look at http://www2.informatik.hu-berlin.de/~tkunze/zaurus/patches/sharpsl_pm-r0.patch ? Apr 13 12:27:43 RP: It tries to move the PXA-specific bits from arch/arm/common to arch/arm/mach-pxa Apr 13 12:28:42 RP: Its not complete yet, but this is how I would do it. Apr 13 12:28:45 koen: angstrom-oabi is softfloat - right? Apr 13 12:29:09 * hrw just removed FPU emu from kernel Apr 13 12:29:11 hrw: for armv4 and armv5, yes Apr 13 12:29:40 still too big Apr 13 12:30:13 Laibsch: Its still a long way until the driver can get into the mainline. the mmc Maintainer doen't like the mmc-spi patch. Apr 13 12:30:36 Laibsch: Some guy on LKML said, that he was rewriting it. Apr 13 12:30:49 thesing: Just wanted to be sure you don't forget to claim it eventually for the work you have done Apr 13 12:30:50 I thought MMC was under a total rewrite anyway to incorporate SDIO stack from MV Apr 13 12:31:21 03hrw 07org.oe.dev * r5433523a... 10/ (4 files in 3 dirs): linux-rp: added SD/MMC driver for collie - written by Thomas Kunze Apr 13 12:31:30 * XorA wonders how much of the bounty people would pay when it hits mainline OE Apr 13 12:31:35 * XorA supects probably most Apr 13 12:33:12 XorA: the sd layer needs to get rewritten to do sdio properly, but the mv one doesn't do sdio properly Apr 13 12:33:23 XorA: You are right. We have to wait and see, if they also incorperate mmc-spi. Apr 13 12:34:52 Laibsch: I hope I remember that, once it gets into the main kernel, in two or three years ;) Apr 13 12:35:50 koen: typical MC, all PR no substance Apr 13 12:42:34 Is there a way to build a gpe-image without having to build task-gpe? Apr 13 12:44:19 what web browser does the nokia 770/800 run? Apr 13 12:45:53 cbrake: by default opera Apr 13 12:47:15 are there any oss browsers in OE that are usable? Apr 13 12:47:58 I see minimo ... Apr 13 12:48:08 s/minimo/megamo/ Apr 13 12:48:22 cbrake: on arm firefox 1.0.7 seems to work Apr 13 12:48:37 koen: thanks, I'll give it a try. Apr 13 12:50:08 koen: on EABI? Apr 13 12:50:30 XorA: I recall using it, but that could be still oabi Apr 13 12:50:41 koen: Ive never got EABI firefox to work Apr 13 12:50:47 OABI 1.0.7 works nicely Apr 13 12:51:02 even when you chroot enough OABI stuff to run it :-) Apr 13 12:51:12 chroot + vnc I hope Apr 13 12:51:17 ;) Apr 13 12:51:48 * XorA begins to wonder if koen has a vnc dildo :-) Apr 13 12:52:24 XorA: no, my the maemo users 'discovered' pocketworkstation.... Apr 13 12:52:45 koen: aaaaaah Apr 13 12:53:07 koen: so they missed the VNC in pocketworkstation was to work around the Qt frontend on the zaurus Apr 13 12:53:22 XorA: sort off Apr 13 12:53:32 XorA: the maemo vkb only pops up in gtk widgets Apr 13 12:53:49 so you need the hacked vnc to get the keyboard to popup in non gtk apps Apr 13 12:56:10 (they could just have written a mb-inputmgr plugin for hildon-panel, though) Apr 13 13:00:43 thesing: Yes, that looks reasonable to me. Can we not reuse the RTAR handling on collie though? Apr 13 13:02:24 I don't know yet. I just wanted to clean things up, before starting some nasty hacks. Apr 13 13:10:37 morning Apr 13 13:21:48 Im being thick today, I can do integer comparison on DAY/MONTH/HOUTH/MIN just by multiply each by 100 to move it out of range of the lesser ones Apr 13 13:22:06 MMDDHHmm (as an int) Apr 13 13:22:08 ? Apr 13 13:22:44 XorA: yes Apr 13 13:23:13 robtaylor: cheers, I was having an insane moment Apr 13 13:23:23 XorA: heh, i know the feeling ;) Apr 13 13:23:39 XorA: it is friday after all :) Apr 13 13:26:03 yes it is very friday Apr 13 13:26:12 13th Apr 13 13:35:49 thesing: That code looks like it should work on a SA1100 too at a quick glance Apr 13 13:38:53 RP you know the issue I was talking about, couldn't figure out why the pcmcia driver didn't work. Well call me stupid, but found out it had been ripped out :D Didn't notice cause all the Kconfigs had changed Apr 13 13:38:57 RP: Maybe. I'll look into this and try. Apr 13 13:45:30 thesing: If there are problem we can factor it out, I'm happy to take patches to do that as long as they're needed Apr 13 13:50:20 RP: http://svn.o-hand.com/repos/sync/ <- does not work :) Apr 13 14:22:53 can some one help me a bit with autoconf and .bb file? Apr 13 14:23:57 how can I pass a username to a SRC_URI using svn+ssh? Apr 13 14:24:20 I tried SRC_URI="svn+ssh://user@blah.com/repository/" but that didn't work Apr 13 14:25:04 * XorA sees more poeple fall over 2066 Apr 13 14:25:50 xuumbi: read packages/ossie/ Apr 13 14:26:58 hrw: ah, so "svn+ssh://user:pass@blah.com/repository/" is the right syntax it looks like Apr 13 14:27:57 wait, my example was wrong, I'm using "svn://user@blah.com/repository;module=mymodule;proto=svn+ssh" Apr 13 14:30:58 SRC_URI = "svn://oe:oe@ossie-dev.mprg.org/repos/ossie/ossie/trunk;module=ossie;p Apr 13 14:30:58 roto=https" Apr 13 14:31:03 works for me Apr 13 14:31:43 Crofton: yes it appears to work with https but not svn+ssh :( Apr 13 14:31:55 ah Apr 13 14:32:01 svnserve with ssh? Apr 13 14:32:10 as opposed to svn with https? Apr 13 14:32:17 what is the reason for getting echo FATAL: 'oe_runconf failed' (http://pastebin.ca/438190). I can't find any configure error in the log. Apr 13 14:32:38 no it's svn over ssh, no svnserve needed Apr 13 14:33:18 I'm creating a bitbake recipe for a build in my company and we use svn+ssh and https Apr 13 14:34:17 xuumbi: hmm, you might need to export SVN_SSH="" Apr 13 14:34:36 probably even before calling bitbake Apr 13 14:35:02 zecke: ah, of course, let me try... Apr 13 14:35:50 xuumbi: http://svn.berlios.de/viewcvs/bitbake/branches/bitbake-1.8/lib/bb/fetch/svn.py?rev=784&view=markup hmm Apr 13 14:36:47 xuumbi: yes we don't export SVN_SSH (even if you would set it in the bb file) Apr 13 14:39:07 zecke: sync is in misc svn atm :/ Apr 13 14:39:55 RP: misc is open? Apr 13 14:41:23 zecke: http://svn.o-hand.com/view/misc/trunk/sync/ Apr 13 14:43:31 thank you Apr 13 14:45:38 zecke: it looks like that worked, thanks! Apr 13 14:46:49 xuumbi: feel free to send a patch to bitbake (file above) to export SVN_SSH to username Apr 13 14:47:43 zecke: ok thanks Apr 13 14:48:05 xuumbi: or at least file a bug report :) Apr 13 14:48:38 Hello Apr 13 14:51:41 hey sirfred Apr 13 14:51:53 sirfred: I used Ximageon without problems today on w3220! Apr 13 14:52:20 then I hit 'rotate' because I didn't know it was Ximageon Apr 13 14:52:39 koen: And that didn't like your w3220 Apr 13 14:52:50 it flipped 180 degrees Apr 13 14:52:59 when I hit rotate again it froze Apr 13 14:53:14 koen: Curious Apr 13 14:53:22 I killed Ximageon and started Xw100 and got the corruption I showed you Apr 13 14:53:38 Because the framebuffer got rotated. that's the cause Apr 13 14:53:54 The problem is that the framebuffer layer doesn't rotate correctly, and the screen gets corrupted. Apr 13 14:55:02 I restored it using fbset, well, actually a program I wrote to test the FBIOPUT_VSCREENINFO ioctl Apr 13 14:55:47 I wonder why there're libxine-x11 and libxine, both are configuring the XShm driver Apr 13 14:57:07 I think I'm going with libxine_1.10.b as version is higher. Apr 13 14:57:14 1.1.0 Apr 13 14:58:22 I'm going to give a try to libxine in the quest for a decent video player for my zaurus Apr 13 14:58:46 koen: do you want me to add #2096 to the Angstrom Meta Bug? Apr 13 14:59:41 hvontres|poodle: go ahead Apr 13 15:00:01 koen: Anyway, I think it's safer to go with Xw100 for w3220, after the changes psokolovsky made to allow offscreen memory usage. Apr 13 15:00:16 sirfred: yes Apr 13 15:00:23 Software rotation should work. Apr 13 15:00:26 sirfred: but today was by accident :) Apr 13 15:00:36 koen: done Apr 13 15:01:00 I only realised it was Ximageon 'killall Xw100' failed Apr 13 15:01:04 koen: It's strange, I was also able to start correctly Ximageon, I had just to comment out a set in ENG_CNTL register. Apr 13 15:01:20 That register should be totally different on W3220 Apr 13 15:02:43 So, it seems that ENG_CNTL writes hung randomly the w3220. Apr 13 15:16:06 http://pokylinux.org/ :) Apr 13 15:16:09 http://xkcd.com/c149.html Apr 13 15:17:04 RP: I'm missing screenshots. ;-) Apr 13 15:17:15 RP: you don't have an nice icons Apr 13 15:18:10 sirfred: http://pimlico-project.org/ has pretty pictures Apr 13 15:18:22 zecke: No. I will mention that... Apr 13 15:18:36 RP: congratulation Apr 13 15:18:41 RP: When it talks about Sato, it says there're "multimedia playback apps", what are they? Apr 13 15:18:45 chouimat: hehe.. xkcd is definitely on my reading list :) Apr 13 15:19:40 Oh, the apps looks great. Apr 13 15:20:04 Too much Matthew Allums everywhere, who know why. ;-) Apr 13 15:20:40 sirfred: Probably http://svn.o-hand.com/view/misc/trunk/owl-video-widget/ Apr 13 15:21:13 RP: Hmm, looking at the names... Apr 13 15:21:35 RP: Is that based in the totem bacon-video-widget? Apr 13 15:21:46 xine or gstreamer based, do you know? Apr 13 15:22:14 sirfred: I'd suspect gstreamer but I really don't know, its not something I've really used Apr 13 15:22:49 Curious, I've been working in the same thing the last two days. I'm trying to make a player using the video widget I've extracted from totem source code Apr 13 15:23:13 Looking at the code, it's gstreamer based Apr 13 15:23:18 sirfred: It has .bb files in poky Apr 13 15:23:27 RP: I'm tempted to try it. :) Apr 13 15:23:34 RP: All this looks great. Apr 13 15:23:53 But the problem is that I've tried bare gst-launch on my zaurus and the performance is far from required. :-( Apr 13 15:24:32 sirfred: Yes, its not good in places :/ Apr 13 15:25:34 Perhaps the 40 bits acumulator in the pxa255 could be used to perform better idct algorithms. Apr 13 15:26:00 s/better/faster Apr 13 15:27:16 RP: I love how it mentions the secret-but-not-so-secret GMAE :) Apr 13 15:27:28 RP: the clyde link 404s Apr 13 15:27:35 What's GMAE ? Apr 13 15:27:47 *shhhhh* Apr 13 15:27:58 :) Apr 13 15:27:58 sirfred: http://www.gnome.org Apr 13 15:28:01 the Gnome mobile and embedded farce^H^H^H^H^H initiative Apr 13 15:28:25 Oh, I didn't know there was such iniciative. Perhaps because it's secret. Apr 13 15:28:35 Is it related with gpe ? Apr 13 15:29:48 sirfred: to some degree Apr 13 15:30:11 RP: "MIT lisence" ? Apr 13 15:30:32 koen: ? Apr 13 15:30:53 poky uses that Apr 13 15:31:26 koen: The same as OE iirc... Apr 13 15:31:53 no, OE use the MIT License Apr 13 15:32:16 koen: ah, you mean the spelling :) Apr 13 15:32:33 :) Apr 13 15:32:34 koen: so RP is a little lysdexic Apr 13 15:32:37 yes :) Apr 13 15:33:02 Lysdexics of the world untie ! Apr 13 15:33:33 Who says RP wrote this? ;-) Apr 13 15:33:59 RP: It says "Richard Purdie" in little characters, at the bottom Apr 13 15:34:32 RP: Have you looked? :-P Apr 13 15:34:58 sirfred: I did :) Apr 13 15:35:03 RP: :) Apr 13 15:37:25 oooooh Apr 13 15:37:26 koen: typo and urls fixed, thanks Apr 13 15:37:35 mallum stole my DPI cluebat Apr 13 15:40:25 Xft.dpi: 100 :-D Apr 13 15:46:37 Seems like there is something wrong with angstrom-mokogateway-image. When i try to build it it gives me this: ERROR: Nothing provides runtime dependency blue-probe Apr 13 15:47:26 isent blue-probe a part of bluez-utils? Apr 13 15:47:45 it's blueprobe, not blue-probe iirc Apr 13 15:48:48 Well then it is a bug somewhere. I have just pasted the ouput Apr 13 15:49:03 maybe in task-base, i'll check Apr 13 15:50:30 seems ok in task-base Apr 13 15:51:59 koen: It's in angstrom-mokogateway-image.bb Apr 13 15:54:25 AM_CONDITIONAL(HAVE_XV, test x$ac_have_xvfoo = "xyes") Apr 13 15:54:27 sabotage Apr 13 15:54:48 It was driving me mad. Apr 13 15:55:21 03koen 07org.oe.dev * r579afa68... 10/ (1 packages/angstrom/angstrom-mokogateway-image.bb): angstrom-mokogateway-image: fix typo and keep up with task-base changes Apr 13 15:55:23 Part of "configure.patch" for libxine Apr 13 15:55:51 looks bogus to me Apr 13 15:55:56 sirfred: libxine patches are AWFUL Apr 13 15:56:26 I was lost because I disabled another patch called no-xv.patch or something so. But there was another trap remaining. Apr 13 15:59:12 I'm not even sure if the result is going to run on the zaurus. Apr 13 16:00:16 With the Xshm video output on zaurus, I got only a black screen under libxine 1.0, the same program was able to play video correctly on x86 Apr 13 16:04:53 cu Apr 13 16:06:27 libxine seems to be unbeatable in alignment warnings. Apr 13 16:07:26 sirfred: dont use libxine, pain will only ensue Apr 13 16:07:47 =) Apr 13 16:07:50 robtaylor: :) Apr 13 16:08:07 robtaylor: it's the last chance. mplayer gui is a hell, gstreamer is slow,... Apr 13 16:08:39 sirfred: gstreamer shouldn't be slow Apr 13 16:08:48 sirfred: what are you trying to do? Apr 13 16:09:01 robtaylor: Playing the same video with mplayer and gst-launch on my zaurus Apr 13 16:09:09 robtaylor: With mplayer: decent, with gstreamer-0.10... well Apr 13 16:09:21 sirfred: what commandline are you using? Apr 13 16:09:36 robtaylor: Let me see... Apr 13 16:10:07 robtaylor: First thing is that I had to disable sound, as the ivorbis plugin couldn't be attached to the pipe Apr 13 16:11:17 sirfred: try gst-lauch-0.10 playbin uri=file:///wheraever/whatever Apr 13 16:11:31 gst-launch filesrc=file.ogm ! oggdemux name=demuxer demuxer. ! queue ! ogmvideoparse ! ffdec_mpeg4 ! xvimagesink Apr 13 16:11:54 robtaylor: Yes, but that way floating point ogg sound decoding drove me to the disaster Apr 13 16:12:07 sirfred: mpeg4 on a zaurus?! Apr 13 16:12:23 robtaylor: Yes. mplayer is able to play that file without dropping frames. Apr 13 16:12:41 Well, I'm lying. It dropped 70 or 80 frames of 12000 or so Apr 13 16:13:08 sirfred: thats pretty impressive. the gst plugin is just using ffmpeg, so it shouldn't be much different Apr 13 16:13:35 robtaylor: Yes, but I don't know where the difference is. Koen said that perhaps the demuxer is the key, iirc Apr 13 16:13:51 That's why I'm now in the xine hell Apr 13 16:13:53 sirfred: what does the playback look like? Apr 13 16:14:09 doesn't gst-ffmpeg contain a copy of ffmpeg? Apr 13 16:14:17 zecke: Yes. Apr 13 16:14:26 sirfred: its plausible that mplay has some arm optimisations that arent swithed on or imported in gst-ffmpeg Apr 13 16:15:12 Could be. But the difference is noticeable. Apr 13 16:15:45 But if I'm not mistaked, ffmpeg is not doing all the job, is it? Apr 13 16:15:47 sirfred: just try with xvimagesink sync=false, let me know if there's a difference Apr 13 16:15:57 sirfred: I don't know how sane the idea is but you could add a queue between file and demuxer :} Apr 13 16:16:18 sirfred: no, as koen says, the demuxer is different Apr 13 16:16:24 Don't know, anyway I had to uninstall all the gstreamer related stuff, I was out of space. Apr 13 16:16:32 :/ Apr 13 16:16:45 And to try libxine spells, I needed size. But I will go back into it after xine failure. Apr 13 16:17:10 sirfred: what size video is this? i might have to investigate where mplayer gets the speed from :) Apr 13 16:17:46 robtaylor: It's 384x256 Apr 13 16:17:53 robtaylor: ogg sound and mpeg4 video Apr 13 16:18:22 robtaylor: If you're interested in the mencoder script... Apr 13 16:18:25 sirfred: and it does mplayer do fixed-point ogg? Apr 13 16:18:37 robtaylor: Yes, tremor based. Apr 13 16:18:47 sirfred: i'll read the source tree ;) is this mplyer svn? Apr 13 16:18:55 robtaylor: The one in bb Apr 13 16:19:18 robtaylor: mplayer_0.0+1.0rc1.bb Apr 13 16:23:38 sirfred: and that was with an xv sink? Apr 13 16:23:44 robtaylor: Yes. Apr 13 16:23:52 robtaylor: mplayer also using -vo xv Apr 13 16:25:41 robtaylor: is gstreamer able to render video without excess memcpy between different elements? Apr 13 16:27:09 ssvb: yep Apr 13 16:27:43 ssvb: though it really depends on the elements used Apr 13 16:28:32 ssvb: the demuxer will probably allocate for each frame though Apr 13 16:29:32 Is there any way to know how much percent of time is spent in any stage of a pipeline (gstreamer) ? Apr 13 16:29:37 ogmvideoparse should just reuse its input buffer, but the main problem will be the buffer allcoation from ffdec_mpeg4 to psas to the videosink Apr 13 16:31:47 robtailor: any extra overhead may kill performance quite easily on arm but modern desktop cpus probably consume more resources in heavyweight codecs such as h246, so probably gstreamer is not too much focused in optimizing other parts of code Apr 13 16:32:57 ssvb: well, the N800 plays mpg4 quite well, but uses a bought-in codec Apr 13 16:33:23 ssvb: instead of ffdec_mpeg4 Apr 13 16:33:51 sirfred: I find that mplayer has a simple design, is easy to hack and has all the necessary capabilities for testing and benchmarking, hard to say that about gstreamer Apr 13 16:34:19 ssvb: Yes, but perhaps lacks some "modularity" Apr 13 16:34:19 hehe Apr 13 16:35:03 zaurus has armv5te, right? Apr 13 16:35:11 robtaylor: Yes Apr 13 16:35:16 robtaylor: pxa255 Apr 13 16:35:36 could just be that gst-ffmpeg isn't being builtwith the opts Apr 13 16:35:54 robtaylor: gstreamer based decoder for mpeg4 on N800 is not so good, it drops a lot of frames on higher resolution videos Apr 13 16:36:02 or it could be excessive malloc Apr 13 16:36:24 ssvb: and mplayer does better? Apr 13 16:37:32 sirfred: mplayer has modularity, but on source level Apr 13 16:38:13 ssvb: well, you see, i use gstreamer for a lot more than just video playback, so they're not equivalent to me Apr 13 16:38:30 sirfred: it is not split into a dozen of binary modules which is not probably good for embedded systems, but avoids dll hell as well :) Apr 13 16:39:17 ssvb: Humm, it just a matter of taste, perhaps, but I was near to get blind after looking at the main() function of mplayer Apr 13 16:39:19 :) Apr 13 16:39:35 ssvb: in *theory~* it should be possible to have gst run as fast as ffmpeg, but noone really seriously does much work with gst-ffmpeg Apr 13 16:40:07 sirfred: same here when trying to debug a problem in mpeg4 decoding.. Apr 13 16:40:26 robtaylor: considering video decoding, mplayer is better on N800 (it is visible on heavy scenes with a lot of panning), but yet inefficient use of Xv still affects performance negatively Apr 13 16:40:29 sirfred: i actually just gave up and worked around it in gst-ffmpeg's wrapper Apr 13 16:41:10 ssvb: yeah, the hardware is 'problematic', to say the least Apr 13 16:41:38 More than 5000 lines of code, full of goto and #ifdefs Apr 13 16:41:45 That's too much for me. Sorry Apr 13 16:43:49 sirfred: I did not look into that horrible code in mplayer (don't touch it and it will probably work :) ), I only worked with video and audio decoding modules Apr 13 16:44:25 Well, my xinelib based player just started, but it says it's not able to handle 'ISO-MPEG4/OpenDivx' Apr 13 16:44:45 At least it's trying to use the xv port Apr 13 16:45:34 robtaylor: the gstreamer tremor thing (gst-plugin-ivorbis) isn't the same as gst-plugin-vorbis Apr 13 16:45:51 dunno why Apr 13 16:46:01 sirfred: oops, not decoding, but output modules, and they seem to contain much less lines of code than gstreamer elements counterparts Apr 13 16:47:32 koen: The problem is that the ivorbis plugin handles different mimetypes than the vorbis one, and cannot be connected to the pipeline. It seems that the ivorbis plugin also handles ogm demultiplexing itself. Apr 13 16:47:48 So, it's unusable to handle a video/audio combined file, iirc Apr 13 16:48:00 indeed Apr 13 16:48:23 But even disabling sound, mplayer run circles around gst-launch Apr 13 16:48:34 hmm, shouldn't be too hard to fix up Apr 13 16:48:55 robtaylor: ivorbis issue or performance? Apr 13 16:49:15 sirfred: ivorbis Apr 13 16:49:39 robtaylor: Perhaps in svn there's some fix for that Apr 13 16:49:46 performance, it'd be intersting to see a log of the gst-ffmpeg build and see if its picking up the arm optimisations correctly Apr 13 16:50:53 robtaylor: Perhaps it's just a matter of tuning or updating the ffmpeg version included with gst-ffmpeg. There was some work lately to improve IDCT decoding, at least for pxa27x and iwmmx Apr 13 16:51:12 sirfred: ah, interesting Apr 13 16:51:56 I also wonder how old is ffmpeg in gst-ffmpeg, afaik ffmpeg does not have formal releases and is only available from SVN Apr 13 16:53:08 ssvb: and it changes API+ABI every 5 minutes Apr 13 16:53:36 the ffmpeg idiots require you to fork it before you can use it Apr 13 16:53:39 koen: Perhaps that's the reason that makes everyone to just use a local copy of ffmpeg. Apr 13 16:53:48 I was curious about that Apr 13 16:53:52 so 'updating ffmpeg' isn't that simple Apr 13 16:54:02 sirfred: IDCT decoding was recently improved for ARMv5 and ARMv6 and an issue preventing IWMMX optimizations from getting enabled was fixed too Apr 13 16:54:22 sirfred: "oooh noooooo! shared libraries are slower than static binaries!!!" is what they seem to say Apr 13 16:54:25 * robtaylor has a problem inidng gst-ffmpeg in oe Apr 13 16:54:37 robtaylor: it isn't in OE yet Apr 13 16:54:40 ah Apr 13 16:54:45 ssvb: Anyway, I'm not in the list of lucky owners of a iwmmx capable chip Apr 13 16:55:00 robtaylor: I opened a bug with a gst-ffmpeg bb stolen from poky Apr 13 16:55:07 aha Apr 13 16:55:23 robtaylor: Let me search the number... Apr 13 16:55:33 koen: probably sticking to some API/ABI and making releases slows down development, that seems to be the only reason Apr 13 16:55:36 also, for gst-ffmpeg the configure is replaced with an autoconf one Apr 13 16:55:41 !oebug 2085 Apr 13 16:55:42 bug 2085 Apr 13 16:55:43 and it doesnt check for arm5v Apr 13 16:55:45 * * Bug 2085, Status: NEW, Created: 2007-04-05 08:41 Apr 13 16:55:46 * * manuel.teira(AT)telefonica.net: New gst-ffmpeg package Apr 13 16:55:47 * * http://bugs.openembedded.org/show_bug.cgi?id=2085 Apr 13 16:55:57 it only checks for IWMMXT and with a TRY_RUN Apr 13 16:56:17 robtaylor: Anyway, it's not the problem for the zaurus. It don't have iwmmx extensions. Apr 13 16:56:28 Only the poor 40 bit acumulator Apr 13 16:56:39 That nobody seems to use Apr 13 16:56:43 sirfred: but there's arm5v optimisations, accoring to standard ffmpeg configure Apr 13 16:56:57 and they're not in the gst replacement configure Apr 13 16:57:06 robtaylor: Interesting Apr 13 16:57:30 robtaylor: armv5 optimizations are quite new Apr 13 16:57:36 If we could have a performance similar to mplayer for gstreamer, the world will be a better place. Apr 13 16:57:52 03manuel.teira 07org.oe.dev * rc65461c5... 10/ (1 packages/gstreamer/gst-ffmpeg_0.10.2.bb): gst-ffmpeg: merge recipe from poky, closes #2085 Apr 13 16:58:11 sirfred: I think Intel used the 40 bit acumulator for IPP and then "forgot" to give us the source :( Apr 13 16:58:11 ssvb: ah, interesting Apr 13 16:58:12 robtaylor: there it is :) Apr 13 16:58:20 koen: heh Apr 13 16:58:29 anyway /me->walk Apr 13 16:58:30 And I would drop my dirty video player prototype and stole poky video widget. Apr 13 16:58:32 laters all Apr 13 16:58:36 robtaylor: cu Apr 13 16:58:55 hvontres|poodle: A pitty Apr 13 16:59:14 hvontres|poodle: Perhaps a decent idct could be implemented using it. Apr 13 17:00:27 koen: Perhaps you want to take a look, talking about gstreamer, to 2057 and 2068 ? Apr 13 17:00:49 sirfred: I doubt that it can be much more efficient than armv5 Apr 13 17:01:54 sirfred: only double MAC is interesting, but it has a high latency and deals only with a single accumulator Apr 13 17:02:10 ssvb: What's double MAC ? Apr 13 17:02:45 sirfred: two multiply&accumulate operations done at once Apr 13 17:03:18 ssvb: ok Apr 13 17:04:09 Have to leave now. cu Apr 13 17:04:14 sirfred: and armv5 can do one 16-bit multiply&accumulate per cycle (that's much faster than armv4 which only has slow 32-bit multiply instructions) Apr 13 17:04:55 03manuel.teira 07org.oe.dev * rb6a0d66e... 10/ (1 packages/gstreamer/gst-plugins-bad_0.10.3.bb): gst-plugins-bad: fix faad2 checks, closes #2057 Apr 13 17:04:59 03manuel.teira 07org.oe.dev * ra212790a... 10/ (1 packages/gstreamer/gst-plugins-ugly_0.10.4.bb): gst-plugins-ugly: fix cross compilatin configure issue, closes #2068 Apr 13 17:38:29 03koen 07org.oe.dev * ra77f3946... 10/ (24 files in 4 dirs): opensync: merge with poky Apr 13 17:41:13 hrw|gone: OE now has a fixed nokia800.conf, you might want to merge it back to poky Apr 13 17:43:49 03koen 07org.oe.dev * r6c22a974... 10/ (1 conf/machine/nokia800.conf): nokia800: add machine from poky, but fix archs and tune spec Apr 13 17:48:03 03koen 07org.oe.dev * r59b29b63... 10/ (6 files in 3 dirs): linux-nokia800: merge with poky Apr 13 18:07:27 koen, I am lazy, do you build Angstrom unstable for EFika? Apr 13 18:09:19 Crofton: http://www.angstrom-distribution.org/unstable/images/efika/20070401/ :) Apr 13 18:09:42 i told you I was lazy Apr 13 18:10:06 Crofton: ask koen to upload the image to your system :) Apr 13 18:10:15 heh Apr 13 18:10:25 I grabbed his OSK image yesterday :) Apr 13 18:10:30 I am not quite that bad Apr 13 18:10:50 I'm, koen subscribes me to the lists I want to be on :} Apr 13 18:11:04 :) Apr 13 18:11:30 sometimes I subscribe people to list they do not want to be on :) Apr 13 18:11:38 zecke: are you going to work ok gtk-webcore or webcore or merge the two? Apr 13 18:11:57 koen: I will be working upstream on the webkit.org project Apr 13 18:12:12 ah, cool Apr 13 18:12:13 koen: I will most likely fix/add/make-compile stuff from the webcore/ directory Apr 13 18:12:36 cmake/scons/autofoo/whatever? Apr 13 18:12:46 koen: I will have to add/fix a buildsystem for Gdk (cmake/qmake?) Apr 13 18:13:05 koen: and then I will need to look at adobe's apollo patches and fix, merge, file them as bug reports Apr 13 18:16:04 RP: I just checked the RTAR Stuff. Its a little bit different from PXA, but some static inline functions in include/asm-arm/hardware/sharpsl_pm.h should do the job. Apr 13 18:16:06 zecke: make sure to get an easy to use gtk interface also. I could not figure out how to use the webkit gdk stuff Apr 13 18:17:51 hi Apr 13 18:18:21 philippe: I think mine will be too easy Apr 13 18:19:53 show_cool_browser() Apr 13 18:20:18 koen: have you looked at apollo? Apr 13 18:20:45 zecke: well you can always take a look at the gtk-webcore interface. It is pretty handy having the renderer as a gtk widget Apr 13 18:21:25 "Apollo is the code name for a cross-operating system runtime being developed by Adobe that allows developers to leverage their existing web development skills (Flash, Flex, HTML, JavaScript, Ajax) to build and deploy rich Internet applications (RIAs) to the desktop." Apr 13 18:21:36 philippe: I have to stack-holder interests. Make my RSS reader use it, propose Chris Blizard to get rid of that mozilla junk :) Apr 13 18:21:53 koen: right, somehow they missed php Apr 13 18:22:25 zecke: getting rid of mozilla would be great. Apr 13 18:22:46 Gecko is next to unusable on my machine. Seems it cannot cope well with multiple cpu's Apr 13 18:22:51 apparently Chriss hofmann convinced people megamo is going to get fixed Apr 13 18:23:08 I don't think OLPC will ditch gecko though :} Apr 13 18:23:22 speically because Chris is on Mozilla Foundation board or such :} Apr 13 18:24:04 koen: what is megamo? minimoz's new name? Apr 13 18:24:21 zecke: I think it is koen slang for minimo Apr 13 18:25:45 unless he is suddenly fond of weird bikes : http://www.megamo.co.uk/home page.htm Apr 13 18:25:57 hehe Apr 13 19:21:00 /quit Apr 13 19:28:12 mickey_bbl: ecore just gained an xcb backend! Apr 13 19:33:48 ah xcb has a kitten logo Apr 13 19:35:24 * koen checks xcb.freedesktop.org Apr 13 19:35:38 zecke: I wonder if glenda likes kittens ;) Apr 13 19:46:33 re Apr 13 19:47:01 sirfred: wb Apr 13 19:49:11 Hm, it seems that the libavcodec bundled with libxine is broken for arm. Apr 13 19:49:20 /usr/lib/xine/plugins/1.1.0/xineplug_decode_ff.so: undefined symbol: put_pixels8_x2_arm Apr 13 19:49:26 Pain and suffering Apr 13 19:54:13 RP: ping Apr 13 19:56:53 koen: Thanks for committing my bugs. Apr 13 19:57:04 Well, the bug fixes, I mean. Apr 13 19:57:10 thanks for fixing them Apr 13 20:00:40 hrw|gone: how can oetest work for you? Apr 13 20:04:24 osama bin undergraduate just closed all the ECE buildings on campus with another bomb threat Apr 13 20:07:07 Crofton: and GGW is coming soon to a virginia tech near you Apr 13 20:07:13 hmm Apr 13 20:07:15 Yes Apr 13 20:07:22 How did you hear about that? Apr 13 20:07:30 #photogeeks Apr 13 20:07:36 heh Apr 13 20:07:40 they for or against? Apr 13 20:07:58 "what lens should I use?" Apr 13 20:08:13 hahahahahaha Apr 13 20:09:28 http://www.petitionspot.com/petitions/ggw-bburg Apr 13 20:09:40 http://www.petitionspot.com/petitions/Girlsgonewild Apr 13 20:09:57 I have actually never seen a ggw video Apr 13 20:10:22 and I've been to the birthplace of ggw: New Orleans Apr 13 20:10:51 File stealling software can get them Apr 13 20:15:01 RP: any idea why maemo.o-hand only has a .deb for dates and not for contacts and tasks? The pimlico website claims the binaries should be there... Apr 13 20:15:19 * koen has a "charge and update nokia tablets" moment Apr 13 20:16:02 Well, I am supposed to report to some place on campus and see what is up Apr 13 20:16:20 of course we all were evacuated b4 the email went out Apr 13 20:16:28 but it is on the way to a bar :) Apr 13 20:20:54 Crofton: Have fun keeping low :) Apr 13 21:04:41 re Apr 13 21:04:56 morning Apr 13 21:05:06 koen|away: opensync svn versions - you did not tested them - right? Apr 13 21:05:42 zecke: I have some stuff added to local.conf iirc Apr 13 21:06:15 hrw: source_checker is not working here, it fetches and fetches :) Apr 13 21:06:46 hrw: so you "upgraded" enough to break it with bitbake-1.6 but not enough to make it work :} Apr 13 21:06:57 zecke: let me check Apr 13 21:08:37 zecke: readin DL_DIR takes eons Apr 13 21:09:02 zecke: but it fetched today Apr 13 21:09:16 I will at least spend another hour on it Apr 13 21:09:25 zecke: and it only fetched new stuff Apr 13 21:09:41 I end up having aalib.tar.gz[1,5] Apr 13 21:10:08 hm. maybe its because of FETCHCOMMANDS which I use Apr 13 21:10:42 I will take a look but even the copy of base.bbclass and bittest did not work at all Apr 13 21:10:51 FETCHCOMMAND_wget = "/usr/bin/env wget -c -t 5 --passive-ftp -P ${DL_DIR} ${URI}" Apr 13 21:12:11 zecke: http://linuxtogo.org/~hrw/{local,site,testrun,sanity}.conf are from my source fetcher dir Apr 13 21:12:31 okay Apr 13 21:12:49 I use bitbake 1.8-svn with bittest Apr 13 21:13:06 http://rafb.net/p/xxJltg71.html now with -c added Apr 13 21:13:11 it is still strange :} Apr 13 21:15:54 zecke: the only change which I have to bitbake there is http://pastebin.ca/438786 Apr 13 21:17:03 ok Apr 13 21:17:07 I go sleep Apr 13 21:17:22 zecke: if you will have some problems/questions feel free to mail me Apr 13 21:17:33 thank you Apr 13 21:18:01 and I thank you - you wrote bittest Apr 13 21:25:09 KOEN! Apr 13 21:29:35 hrw|gone: the way we handle stamps changed :) Apr 13 21:40:14 Good night Apr 13 21:43:33 ouch, I need to ask for the removal of altboot :} Apr 13 21:45:26 zecke: removal from where? Apr 13 21:45:46 OE Apr 13 21:46:08 Why? altboot is destined to be used for the SlugOS replacement ... Apr 13 21:46:29 in contrast to our COPYING file these files are GPL licensed Apr 13 21:47:03 zecke: pong Apr 13 21:47:06 OE individual files can't be GPL'd? Isn't it just mere aggregation? Apr 13 21:47:40 koen|away: No idea about maemo, you'll need to ask someone else at OH, probably on Monday :/ Apr 13 21:48:28 rwhitby: It should really be under one licence or it will get confusing :/ Apr 13 21:48:43 Its at least a very valid question to raise... Apr 13 21:49:06 yep. Apr 13 21:49:08 rwhitby: .bb files can't be under GPL Apr 13 21:49:19 rwhitby: it has been this way since the creation of OE Apr 13 21:49:34 zecke: I don't disagree, but what's the rationale? Apr 13 21:49:43 rwhitby: specially we have argued with CoreDump and this is why I'm schocked to see these files Apr 13 21:50:18 rwhitby: we should stick to one license. we advertise it as such Apr 13 21:50:56 rwhitby: why it has to be MIT? Well I can't comment on schurigs/mickeyls/pb_s/kergoth's decision Apr 13 21:51:13 * rwhitby likes MIT Apr 13 21:51:23 rwhitby: personally I like the MIT license, and we have yet to see a run-a-way fork where GPL would have "helped" Apr 13 21:52:08 I simply wasn't aware that bb files had to be MIT. Apr 13 21:52:53 COPYING.MIT :) Apr 13 21:53:20 zecke: so CoreDump|afk could keep the altboot package as GPL, but just put the .bb file as MIT, right? Apr 13 21:53:37 rwhitby: from my understanding only the patches use the license(s) of the original package Apr 13 21:53:51 rwhitby: right, same a start scripts, etc Apr 13 21:54:05 rwhitby: There is no problem with GPL based packages, this is purely about the data in OE itself Apr 13 21:54:43 zecke: Arguably all my kernel patches are GPL and there are odd ones in OE ;-) Apr 13 21:55:02 I don't want to end up with: "OE is MIT*" Apr 13 21:55:16 zecke: perhaps there should be a Policies link on the left sidebar of www.openembedded.org ... Apr 13 21:55:19 *)besides the two methods in base.bbclass, conf/*, :) Apr 13 21:56:16 and perhaps the important fact of BB licensing should be somewhere on http://www.openembedded.org/wiki/StyleGuide, ahead of whitespace conventions :-) Apr 13 21:56:41 We suck at documentation :/ Apr 13 21:57:32 we should save money to get a doc writer... Apr 13 21:58:38 I don't even see a 'register' link on the OE wiki so I could add it myself ... Apr 13 21:58:39 rwhitby: ironically I know more people/companies that "stole" Opie than stole OE Apr 13 21:58:57 rwhitby: you can edit pages without being registered Apr 13 21:59:04 rwhitby: you can ask koen for an account Apr 13 22:00:19 * rwhitby anonymously adds "All .bb files should be licensed under the Microsoft EULA." to the StyleGuide ... Apr 13 22:01:20 rwhitby: :) Apr 13 22:01:29 OE should migrate wiki to decent app, away from that D.., D..., forget D-what Apr 13 22:02:23 how about "All .bb files must be relesed under the MIT license. This restriction does not apply to the source code of, or patches to, the packages which are built - it only applies to the recipes (and associated non-patch files) in the OE metadata repository." Apr 13 22:02:47 ack Apr 13 22:02:58 this documents what I have always believed Apr 13 22:03:01 http://www.openembedded.org/styleguide Apr 13 22:03:08 psokolovsky: drupal, drupal, drupal Apr 13 22:03:23 ~lart drupal ;-) Apr 13 22:03:24 * ibot overclocks drupal ;-) until drupal ;-) burns out Apr 13 22:03:25 ack from me too. It is what historically has been laid down Apr 13 22:03:46 psokolovsky: have you tried to contact TT? Apr 13 22:04:06 zecke: regarding what? Apr 13 22:04:25 psokolovsky: Qtopia :) Apr 13 22:04:31 lol http://qtopia.net/modules/devices/ Apr 13 22:04:49 "Adaptor Alert Notice: Universal adaptor malfunction alert" and you need to login to know if your adapter explodes :} Apr 13 22:06:20 well, I don't do qtopia, i do opie, a bit. Apr 13 22:07:21 psokolovsky: well, don't you want to do Qtopia4? Apr 13 22:07:49 psokolovsky: Did you ever figure out a solution to the circular dependancy with OpieMediaPlayer1 in .dev? Apr 13 22:09:00 zecke: likely, no ;-I I'm interested to polish sth already existing and working, and trying my "generic device support" ideas, not building something again from half-done state Apr 13 22:09:32 hvontres|poodle: I thought I had, but somebody told me they're there again for bb1.8 ;-( Apr 13 22:09:43 hvontres|poodle: consider submitting bug with logs %) Apr 13 22:10:23 I've looked at theose circular depends and I think bb 1.8 is actually correct :/ Apr 13 22:11:02 psokolovsky: ok. I wanted to see if you had something first. I'll try to get you a buggy log. for now I commented mediaplayer1 out of task-opie Apr 13 22:11:26 Ah crap, I see the issue, hope to fix next few days Apr 13 22:11:38 RP: yep, seems I didn't finish it Apr 13 22:12:13 psokolovsky: so do you still want those logs? Apr 13 22:12:29 wb mickeyl Apr 13 22:12:36 thanx Apr 13 22:13:00 hvontres|poodle: yep, if possible, because I don't run 1.8 still Apr 13 22:13:31 psokolovsky: Any particular reason why out of interest? Apr 13 22:13:57 psokolovsky: np. I'll be home in about 2.5 hrs. Apr 13 22:15:09 RP: in interest, but I need to resize my OE partition to accomodate all that extra stuff it will build due to Debian renaming ;-) Apr 13 22:15:19 RP: is there a way to get more debugging output from the serial console? right now all I get is the standard console output.. Apr 13 22:15:32 hvontres|poodle: thanks Apr 13 22:17:31 hvontres|poodle: Change the kernel log level Apr 13 22:18:11 psokolovsky: Right. I wish it didn't do that tbh but it is correct :/ Apr 13 22:18:46 Drop Debian renaming from Angstrom! Drop renaming! ;-D Apr 13 22:20:15 psokolovsky: Or add a cache Apr 13 22:21:27 regarding MIT. when we started OpenEmbedded we looked out for the most liberal (including, but not limited to commercial usage) open source license and to our understanding that was MIT. Apr 13 22:22:19 * RP thinks it was a good choice for OE Apr 13 22:28:19 zecke: actually, that adapter was a freebee, and the problem is that when the plugs are in, they are still live, so if you happen to touch them when it is still plugged in... which has nothing to do with Qtopia, or the greenphone Apr 13 22:29:04 ljp: :) Apr 13 22:29:08 ljp: how are you? Apr 13 22:29:15 am ok. Apr 13 22:29:21 good Apr 13 22:29:28 i actually got oe to build complete Apr 13 22:29:36 hehe :) Apr 13 22:29:44 am retrying to see if it was just a fluke Apr 13 22:29:53 ljp: need a hand on building QtopiaCore? Apr 13 22:30:08 do you know if the iwmmx compiler builds? and works Apr 13 22:30:38 not building qtopiacore, just console, at the moment Apr 13 22:30:41 ljp: which one? Apr 13 22:31:09 anyone, we have had problems building iwmmx toolchain Apr 13 22:31:36 ljp: I know Sean Kellys mail Apr 13 22:31:45 ljp: I have a special moto compatible gcc3.3 toolchain Apr 13 22:31:55 ljp: gcc3.4,4.1 should work out of the box Apr 13 22:32:04 need 4.1.1 at least :) Apr 13 22:32:21 for symbol visibility Apr 13 22:33:11 ljp: by changing the tuning to march=iwmmxt you should get a building one Apr 13 22:33:48 ljp: There are visibility patches for 3.x available and working in OE iirc Apr 13 22:34:35 would like to stick with gcc 4 Apr 13 22:35:26 ljp: I've used OE's 4.1.1 with visibility and iwmmx... Apr 13 22:35:49 zecke : have you done anything with the greenphone? Apr 13 22:36:13 anyway to specify where the toolchain lives? Apr 13 22:36:36 ljp: yes and no. I'm still struggling with the serial and I have a half complete version of the OpenMoko rss reader for Qtopia Apr 13 22:37:32 ljp: which toolchain? the one you are going to build? Apr 13 22:37:38 ljp: or if you want to use an external one? Apr 13 22:37:44 yes. oe built one Apr 13 22:38:02 ljp: yes, check conf/bitbake.conf Apr 13 22:38:06 no, we need to be able to install the toolchain in other machines - teambuilder farm Apr 13 22:38:33 ljp: a) search for ${TMPDIR}/cross in bitbake.conf Apr 13 22:39:14 can I override that from local.conf? Apr 13 22:39:54 ljp: yes Apr 13 22:47:16 03pfalcon 07org.oe.dev * r04adb86e... 10/ (13 files in 5 dirs): Apr 13 22:47:16 opie-mediaplayer1: Finish clearing loops in dependencies. Apr 13 22:47:16 * Plugins now depend on opie-mediaplayer1-core, which is core in itself. Apr 13 22:47:16 * opie-mediaplayer1 is now a virtual package, just collecting core and Apr 13 22:47:16 plugins together. Apr 13 22:47:17 * So, no loops. Apr 13 22:47:44 tried to get yuhuatel to open their 17 binary modules... no go Apr 13 22:48:27 so gp has the destinction as having the most binary modules of any linux embedded device Apr 13 22:48:45 psokolovsky: I'll test before and after updating the db then. Apr 13 22:49:19 hvontres|poodle: feel free to test only after ;-). I'm confident that should solve it ;-) Apr 13 22:49:40 psokolovsky: np. Apr 13 22:49:45 ljp: hehe, have you seen the neo qemu port Apr 13 22:49:58 not yet Apr 13 22:50:01 ljp: you could now easily create a qemu board file for the gp Apr 13 22:50:14 ljp: running something close to the gp kernel Apr 13 22:50:54 will look into it Apr 14 00:24:23 * rwhitby tries OE build on RHEL5 ... Apr 14 00:57:11 nope, can't easily get 'git' for CentOS 5 (RHEL5) yet. **** ENDING LOGGING AT Sat Apr 14 02:59:56 2007