**** BEGIN LOGGING AT Sat Jan 10 02:59:57 2009 Jan 10 08:57:20 Ainulindale: thanks, I'm a git newbe :-p my first git push ever Jan 10 09:07:42 mw|: take a look at gitk and git gui... both included in git and *very* useful :-) Jan 10 10:00:04 Hire: buon giorno :-) Jan 10 10:01:11 guten morgen mrmoku Jan 10 10:01:23 ( i know only this two words :D ) Jan 10 10:01:35 *those Jan 10 10:02:26 hehe, I'm sure you also know the word Bier ;) Jan 10 10:03:40 sure :D Jan 10 10:04:48 und Bretzel! Jan 10 10:07:29 Hi guys! Jan 10 10:07:35 ptitjes: buon jours Jan 10 10:21:41 wow Jan 10 10:21:47 so many diff Jan 10 10:21:54 and no a new image for SHR? Jan 10 10:22:56 mrmoku: quite good! bonjour! Jan 10 10:23:44 nothing is better than internationnal collaboration to learn languages ;) Jan 10 10:24:08 another analcholic french? Jan 10 10:24:52 ptitjes: or learn english Jan 10 10:25:02 whihc seems to be the language of international collaboration Jan 10 10:25:03 :/ Jan 10 10:25:13 héhé Jan 10 10:26:46 raster: hey, how's life in list-world? Jan 10 10:27:05 fine Jan 10 10:27:06 list works Jan 10 10:27:25 was writign the wiki elementary intro Jan 10 10:28:26 oh raster will you have something list with text entry and list comprehension ? Jan 10 10:28:50 ????? Jan 10 10:29:01 raster: uhh, so I have no excuse anylonger to start coding an Elementary app... :-) Jan 10 10:29:08 you lost me there Jan 10 10:29:11 say completion Jan 10 10:29:17 mrmoku: hahaha Jan 10 10:29:31 ptitjes: completion? Jan 10 10:29:36 still don't get it Jan 10 10:29:50 mrmoku: raster is on the bed now... mallc instead works on elementary Jan 10 10:30:02 you know: you type some letters and this filters the list Jan 10 10:30:20 and I prefer so. malloc is more accurate than raster and doesn't messed up E every 20 commits Jan 10 10:31:24 ( :x ) Jan 10 10:31:52 ptitjes: oooh - no. no such thing there. its a simple list of items with labels and widgets u can place before/after the label Jan 10 10:32:40 raster this is nice either :) Jan 10 10:33:58 raster: I read some days ago about your choice for contacts/messages file backend, etc... Jan 10 10:34:51 however will you have ipc between your apps to regulate their displaiy ? Jan 10 10:36:01 let me explain : for instance you may want new incomming messages to display just with a simple notification when you are in an active call Jan 10 10:36:28 or priority stuff like that... Jan 10 10:36:31 ? Jan 10 10:43:22 can just use file locking Jan 10 10:44:12 no context or activity sharing ? Jan 10 10:44:25 ptitjes: you can put a number on the messages icon Jan 10 10:44:27 no Jan 10 10:44:33 why? Jan 10 10:44:44 messgaes or contacts are a database effectively Jan 10 10:44:53 all you need is to get the data Jan 10 10:44:54 modify it Jan 10 10:44:57 and know when its changed Jan 10 10:45:04 knowing when is as simple as d/inotify Jan 10 10:45:15 modification is simple file locking Jan 10 10:45:42 even then it isnt needed that often Jan 10 10:46:55 For my use case for instance, if you don't have a context protocol, your dialer app has to know about messages Jan 10 10:47:23 This isn't much a clean way Jan 10 10:48:36 it can just load them up like anything else Jan 10 10:49:24 i dont intend to change anything because what i have works Jan 10 10:49:29 enough for me and my needs Jan 10 10:49:58 changing it will just make development harder/slower and waste time Jan 10 10:49:59 raster: I was just asking to know your view on this Jan 10 10:50:10 there is no request Jan 10 10:50:14 :) Jan 10 10:50:19 i like the files Jan 10 10:50:21 its simple Jan 10 10:50:26 easy to backup and restore Jan 10 10:50:28 I do too Jan 10 10:50:33 easy to import/export Jan 10 10:50:48 watching for changes is easy - watch the filesystem Jan 10 10:50:55 making changes is easy Jan 10 10:50:59 loading data is easy Jan 10 10:51:16 everythng pretty much revovles around that Jan 10 10:51:45 I just feel that, if someone wants to make a companion app along to your apps, there must be a protocol between this new app and yours to negociate the real screen estate Jan 10 10:52:00 oh Jan 10 10:52:05 thats what a windowmanager is for Jan 10 10:52:16 managing the data is one thing Jan 10 10:52:17 window manager is for windows Jan 10 10:52:19 managing the screen is another Jan 10 10:52:22 yes Jan 10 10:52:24 aren't there any shr developments anymore? or why does opkg update && opkg upgrade do nothing since weeks Jan 10 10:52:27 and the other app has - other windows Jan 10 10:52:47 app-specific (like phone matters) context manager is another thing Jan 10 10:53:15 tifereth: lots of activity... no new builds though Jan 10 10:53:15 tifereth: do you use testing or unstable Jan 10 10:53:20 should change soon though Jan 10 10:53:28 testing i thing Jan 10 10:53:31 think Jan 10 10:53:32 tifereth: most of the dev activity is on unstable... Jan 10 10:53:41 ptitjes: if you have a notification - you pop up a window Jan 10 10:53:45 that will appear Jan 10 10:53:49 pretty simple Jan 10 10:53:49 :) Jan 10 10:54:13 does the "no builds" thing lead to the git problem? Jan 10 10:54:21 raster: but if your are in an active call, one may want that this doesn't get disturb... do you see my point ? Jan 10 10:54:54 that is one of the things unresolved in a phone "desktop" Jan 10 10:54:57 you may not Jan 10 10:55:03 but if you want to notify and have something to notify Jan 10 10:55:10 it will appear on top Jan 10 10:55:24 when it goes away (you dismiss it) it will go away and you go back to where you were Jan 10 10:55:29 tifereth: what git problem? Jan 10 10:55:34 as such there is nothing mroe sophisitcated than that Jan 10 10:55:59 i havent had time to do it - at openmoko i spent to much time trying to do a pixel-perfect implementation of will/weans flash designs Jan 10 10:56:14 mrmoku: http://lists.openmoko.org/pipermail/devel/2009-January/003987.html this one Jan 10 10:56:17 and arguing ab out qwerty control keys and the vkbd dictionary stuff than actualyl solving real core problems Jan 10 10:56:28 and this is still hapening if i want to build locally Jan 10 10:56:30 :) Jan 10 10:56:38 things like using qtopia brought extra complicatiosn thansk to qtopias softmenu window at the bottom Jan 10 10:56:47 as it assumes it will have this on the display Jan 10 10:56:57 so i havent solved many of these things Jan 10 10:57:06 vkbd protocol still is not totally fixed either Jan 10 10:57:12 only aa limited set of kbd-type hints Jan 10 10:57:18 raster: so would you be for a context/activity manager thing just as there is a window manager ? Jan 10 10:57:19 no way to separately hint "dict fixingon/off" Jan 10 10:57:26 and any way to hint for specific languages/dicts Jan 10 10:57:35 you dont need that Jan 10 10:57:40 you need the window manager only Jan 10 10:57:44 the rest is hints Jan 10 10:57:51 and window management policy Jan 10 10:58:29 eg an app can just set a hint on the window (property) advising "do not interrupt me ever" Jan 10 10:58:40 and so the wm will queue up or hide any windows that would normally interrupt it Jan 10 10:58:49 the wm can get windows tha are "information dialogs" Jan 10 10:58:58 and specialyl manage them (put them at the bottom of the screen in a stack) Jan 10 10:59:01 etc. etc. Jan 10 10:59:03 raster: the "do not interrupt me ever" hint is not extensible Jan 10 10:59:12 just with the right hints the wm then knows what to do - and can do anything it likes Jan 10 10:59:25 all hints are extensible Jan 10 10:59:28 by definition Jan 10 10:59:30 what if one writes an application that has not been anticipated by the one that sets an "do not interrupt me ever" hint Jan 10 10:59:44 tifereth: hmm... I did a complete build two days ago... though I'm not getting the SHR stuff via git/http Jan 10 10:59:47 trying that now Jan 10 10:59:48 and that this application should interrupt the other in the user point of view ? Jan 10 10:59:53 you write apps within the hints available Jan 10 11:00:03 if you dont like it - you live with it Jan 10 11:00:10 or you lobby for more hints and make your case Jan 10 11:00:10 that would be great, due i'm not sure where the issues come frome Jan 10 11:00:24 even if i get a step further some times ago when i tried to build Jan 10 11:00:34 ptitjes: then u have a window priority Jan 10 11:00:46 windows of priority X never get interrupted by windows of priroty < X Jan 10 11:00:49 as i said Jan 10 11:00:50 raster: this is the current phone providers way of putting thing that are not nice for users Jan 10 11:00:53 it can all be done with hints Jan 10 11:00:59 but we arent in a position to do this currently Jan 10 11:01:03 tifereth: hmm, cloning the shr repos via http works fine for me... what git version are you using? Jan 10 11:01:19 this is too much statical Jan 10 11:01:30 s/al// Jan 10 11:01:30 ptitjes meant: this is too much static Jan 10 11:03:37 how is it different to a context manager? Jan 10 11:03:45 the window manager ACTS as context manager Jan 10 11:03:48 that is the job of a wm Jan 10 11:03:53 to handle window focus and display Jan 10 11:04:08 and arrangement Jan 10 11:04:08 it implements some for of policy Jan 10 11:04:11 and generally also provides config options for users to tune than policy Jan 10 11:05:20 as such i dont like the general direction of illume as it basically isnt doing a lot of the ui things i had expected/hoped it should Jan 10 11:05:30 i was pretty limited in how to do it Jan 10 11:05:46 in fact the android/plam pre ui's are much closer to what i was planning and going to do Jan 10 11:06:05 Say you may want to add soft rules like : if I'm dialing a person X and that person X is dialing me at the same time then "dismiss my outgoing call if X is my boss or dismiss X's incomming call if X is a friend (so that I can be charged for the call)" Jan 10 11:06:14 (minus the compositing and scaling of the window live) Jan 10 11:06:51 dude something like that is plain oimpossible to do without forcing users to write code to do it Jan 10 11:07:03 dont even try to solve things at that level of detail Jan 10 11:07:07 stand back and do it generically Jan 10 11:07:10 raster: this is where I disagree Jan 10 11:07:25 I want user to able to such express rules graphically Jan 10 11:07:33 mrmoku: git version 1.5.4.3 Jan 10 11:07:53 raster: yeah generically means a phone-stuff specific context manager Jan 10 11:07:59 tifereth: you could also try to rm -rf downloads/git_shr* in your toplevel builddir Jan 10 11:08:21 tifereth: or you could upgrade git and see if that works... Jan 10 11:08:42 though I think it should also work with an older git Jan 10 11:09:08 ptitjes: and when the app interrupting is not a phone app Jan 10 11:09:13 its gps tellign you of a wifi spot nearby Jan 10 11:09:19 okay i'll try Jan 10 11:09:24 or your irc app sayoung you have a mprivate message Jan 10 11:09:32 or firefox saying "download complete"? Jan 10 11:09:35 stand back Jan 10 11:09:40 and look at the problem much mroe generically Jan 10 11:10:32 mrmoku: rm didn't work so i'll try to upgrade git now Jan 10 11:12:13 ok Jan 10 11:12:47 freesmartphone.org: 03jluebbe 07framework * r1b882a8b5f87 10/framework/subsystems/ogpsd/factory.py: ogpsd: use framework.config like ogsmd.device Jan 10 11:17:03 freesmartphone.org: 03jluebbe 07framework * r052a4b4adac2 10/framework/subsystems/odeviced/kernel26.py: odeviced: readd fcntl import which got lost during the conversion to KObjectDispatcher Jan 10 11:20:45 Ainulindale: ping, I have some questions about lfg, phonegui, branching and how to go on Jan 10 11:23:14 hi mickey|zzZZzz, is gsm0710muxd 0.9.2.2 still compatible with frameworkd from the stabilization branch? Jan 10 11:27:16 mrmoku: O_o i upgraded to the latest git and still the same error Jan 10 11:27:37 i'll start from the beginning one more and then i'll give up Jan 10 11:29:17 quickdev: hey... how's web-programming-life... boring isn't it ;) Jan 10 11:30:02 tifereth: you could try to just clone the shr repo in a tmpdir... see if that works Jan 10 11:30:12 git clone http://shr.bearstech.com/repo/shr.git Jan 10 11:30:27 okay i'll try Jan 10 11:30:47 mrmoku, sometimes it's boring, but not if you redesign certain things. I have to improve my page, because at the moment I'm spending more money for advertising than money comes in ;) Jan 10 11:31:53 quickdev: problem with designing is... depends on personal preferences... that's why I gave up weg-programming a long time ago. And because of browser-differences in that days Jan 10 11:32:14 hi quickdev Jan 10 11:33:30 quickdev: I did a stabilization/milestone4 branch in lfg (though I don't know if we still need it)... how do you want to handle it in phonegui-efl? Jan 10 11:39:11 mrmoku: the clone thing works, but i'm not sure if that is what will be executed with make image and produces the error Jan 10 11:39:57 tifereth: hmm.. should be though... Jan 10 11:56:49 hello Jan 10 11:56:49 nomeata: There were no API changes. Just bug-fixing (not and ideal solution, i must admit, but working). Jan 10 11:57:02 PaulFertser: thx Jan 10 11:59:07 nomeata: Thanks for you Debian efforts. In my opinion Debian is much more user-friendly comparing to OE, i hope it'll become a preferred choice for majority of "ordinary users" when the things will stabilize. Jan 10 11:59:27 I’d appreciate it :-) Jan 10 11:59:53 But I also have to acknowledge that OE is more suitable for auto-building, which means less work to package new upstream versions Jan 10 12:00:09 (just change a variable, instead having to download, package, build, tag, upload it manually) Jan 10 12:03:05 Someone should come with a solution to automate it at least for common cases of making packages for develepment snapshots (at least for git and svn). It doesn't look to be too hard, a common bash scripting will suffice (in my uneducated guess). Jan 10 12:29:08 morning folks Jan 10 12:29:18 mornin' Jan 10 12:31:46 big m! Jan 10 12:35:10 mickeyl: Morning. Thanks for the kind words. :) A non-related question: Are you sure that "sending reinit in 5 seconds.." and "timeout_add_seconds( 10," match? Jan 10 12:35:32 ya, didn't bother to change the text :) Jan 10 12:35:43 will fix it next time i touch that file Jan 10 12:37:54 i'm very glad at+copn works Jan 10 12:38:08 i wasn't sure whether this was a bug on the firmware muxer or on our side Jan 10 12:38:15 since it always worked in single mode Jan 10 12:38:20 but got corrupted in muxing mode Jan 10 12:40:28 mickeyl: i'm going to try a more elegant solution (i think i can add a Glib iowatch on "write will not block" condition and flush buffer there, instead of doing it in a watchdog). And, btw, i tried to reproduce the bug mato mentioned: Jan 10 12:40:45 good Jan 10 12:40:51 i'd also welcome the async approach Jan 10 12:43:02 open mickeyterm, it works, close it. Use mdbus to get a pty, do "cat /dev/pts/1", ^C it, open mickeyterm again. Sometimes a channel won't be opened, but it works next time. I haven't investigated it thourougly yet, but i've never seen an issue with plain mdbus, cat (without using mickeyterm). Jan 10 12:43:53 mickeyl: i'll try to implement async approach channel allocation approach today, but no promises ;) Jan 10 12:43:53 sure, cool Jan 10 12:50:13 Does anyone have any idea why using nandwrite to flash u-boot on running FR can fail? One guy bricked his GTA01 using it (not that anyone even suggested to do that to him). Jan 10 12:52:24 not exactly an expert wrt. nand, but is nandwrite honoring bad blocks by default? Jan 10 12:53:41 I'm investigating that atm. I'm not sure badblocks matter for a bootloader area. Jan 10 12:56:21 they do, we only have the guarantee of about the first 1K blocks to be good Jan 10 12:56:45 that's why Laf0rge invested some good time to come up with the bad block stuff in U-Boot Jan 10 13:09:24 heyho Jan 10 13:09:36 moin Jan 10 13:10:10 mickeyl: thanks, that you added python-elementary to oe master Jan 10 13:10:39 np Jan 10 13:10:49 it still doesn't work for me, but hey... Jan 10 13:10:53 *shrug* :) Jan 10 13:11:07 :) Jan 10 13:11:18 what doesn't work? Jan 10 13:12:33 still the SIGSEGV when i do anything Jan 10 13:13:23 maybe e libs are too old? Jan 10 13:13:37 i always update everything Jan 10 13:13:46 [as in ... actually compiling] Jan 10 13:13:47 that actually causes the most problems Jan 10 13:13:54 hm Jan 10 13:14:14 don't worry, must be something local since it works for you Jan 10 13:14:28 and for many others :) Jan 10 13:16:30 i had high hopes for elementary's list widget, but it's pretty slow on our hardware Jan 10 13:16:44 not a pun on elementary, it's for sure the hw, still kind of annoying Jan 10 13:16:45 on the fr? Jan 10 13:16:48 yes Jan 10 13:16:56 thats not good Jan 10 13:17:12 i'll check on other hardware soon Jan 10 13:17:13 I am currently working on including the list widget in python-elementary Jan 10 13:17:21 good Jan 10 13:17:23 it's a good widget Jan 10 13:17:26 jep Jan 10 13:17:47 ooh! Jan 10 13:17:54 first time no SIGSEGV Jan 10 13:18:14 somewhere in the elementary sources I read that raster want to improve the internal structure of the stored list items a bit to speed up Jan 10 13:19:39 morphis: that will come with the gurana merge Jan 10 13:19:42 it has a mvc list Jan 10 13:19:49 thats intede for long lists of lots of items Jan 10 13:19:53 the current one is for short lists Jan 10 13:19:56 but is more flexible Jan 10 13:20:07 raster: ah ok Jan 10 13:20:47 hi folks Jan 10 13:20:48 raster: the merge with gurana will bring only new widgets or event the whole mvc structure? Jan 10 13:20:55 hey bumbl Jan 10 13:20:57 new widgets Jan 10 13:20:57 morphis: have you fixed the button segfault? Jan 10 13:21:02 change some widget core infra you dont see Jan 10 13:21:09 bumbl: the toolbar button segfault? Jan 10 13:21:10 and improve keyboard control and focus handling Jan 10 13:21:17 raster: fine Jan 10 13:21:25 but elm has many more widgets than grn at the moment Jan 10 13:21:36 bumbl: toolbar is now complete Jan 10 13:22:00 morphis: the segfault concerning the test app and shr settings - when pressing a button it segfaults Jan 10 13:22:38 (the window which should be opened segfaults the test app stays alive) Jan 10 13:22:51 morphis, have you fixed the free bug? maybe that's it.. Jan 10 13:23:06 quickdev: the free bug is fixed in current svn Jan 10 13:23:13 ah, ok :) Jan 10 13:24:03 bumbl: with the current version I have no more segfaults somewhere Jan 10 13:24:09 * morphis will be back in some minutes Jan 10 13:24:25 i try to update again (last update yesterday) Jan 10 13:24:44 i was not alone though Jan 10 13:24:51 dos had the same problems Jan 10 13:31:09 :/ linux-openmoko faild to build on Hunk #1 Jan 10 13:32:17 yoyo: ya Jan 10 13:32:26 working on that Jan 10 13:32:51 mickeyl: it will work (r1) on gta01 ? (r0) didn't start Jan 10 13:33:16 didn't try that since a while Jan 10 13:33:42 we'll release fso5 for 01 as well though, so we need to bump it into shape Jan 10 13:33:55 yesturday I try at almoust all posible ways witout dboard Jan 10 13:35:07 bumbl: could you give me a backtrace again, please? Jan 10 13:36:22 morphis: yes - updating at the moment - then i will have a look if the problem still exists Jan 10 13:37:39 bumbl: ok Jan 10 13:39:16 moprhis: with latest python-elementary i have segfaults on test.py too Jan 10 13:40:08 where do you clicked in the test app? Jan 10 13:40:27 morphis: Bg Plain ;p Jan 10 13:40:31 Bg Image too Jan 10 13:40:33 on the button (regardless which one) Jan 10 13:40:42 on the fr or on local? Jan 10 13:40:45 Icon Transparent too Jan 10 13:41:13 morphis: i have it on FR Jan 10 13:41:48 dos1: ok, on my local pc I never get with current svn a segfault Jan 10 13:42:22 morphis: do you have a shr-unstable on freerunner? can you test it? Jan 10 13:43:45 dos1: yes I have, flashed yesterday Jan 10 13:43:55 I will try it later Jan 10 13:43:59 hello Jan 10 13:44:11 hi Jan 10 13:49:28 mickeyl: Qi definetely checks for bad blocks while reading the rest of itself from NAND. But there's a difference between GTA01 and GTA02, in GTA01 bad block information is stored OOB (and pagesize is 512) and on GTA02 bad block information is stored (as far as i can understand) in-band and block size is 2048. So tests, confirming that nandwrite works on GTA02 has nothing to do with GTA01. nandwrite seem to support badblocks Jan 10 14:10:04 mickeyl: to me the kernel code seems to be corresponding to what a bootloader expects. I'll try to flash Qi from running system on my FR nevertheless. Jan 10 14:10:28 oki Jan 10 14:10:30 bbiab Jan 10 14:10:59 PaulFertser: for testing different kernel versions automatically i'd need some sort of fallback support from bootloader Jan 10 14:11:06 PaulFertser: any ideas how this could be done efficiently? Jan 10 14:11:50 PaulFertser: could uboot 'saveenv' a fallback kernel to boot before booting the kernel? Jan 10 14:16:10 lindi-: i recommend to perform all this controlled from a PC. You can load a kernel to memory and try to run it. If you can get ssh to the phone in some 120 sec, e.g., that kernel is good. If not, you might have to reset FR (that probably can be done with the help of DB or with a PC-controlled switch of battery and usb power. If the kernel is stuck, you have to somehow reset the device anyway. Jan 10 14:16:41 PaulFertser: ah, load kernel to memory? did not even think about that Jan 10 14:16:49 PaulFertser: wonderful Jan 10 14:17:22 lindi-: you can do it with dfu-util even. Jan 10 14:17:34 PaulFertser: do you know if openmoko inc. has some automated tests like these? Jan 10 14:18:16 lindi-: or you can setup u-boot to unconditinally load kernel to memory from tftp server you can easily control. Jan 10 14:19:15 lindi-: have no idea about OM-the-company. I'm not sure it makes much sense to perform automatic kernel testing especially if you have only one device to care about (sadly, but OM doesn't care much these days about GTA01). Jan 10 14:19:45 PaulFertser: ok, i'm thinking about doing much more than just checking that the kernel boots Jan 10 14:19:58 PaulFertser: wlan/suspend stress-testing at least Jan 10 14:20:45 PaulFertser: i did not even know uboot can do TFTP Jan 10 14:20:51 Try to include resuming initiated by call in your tests ;) I think there might be some problems sometimes (i couldn't manage to dive that deep yet to confirm it)> Jan 10 14:21:25 PaulFertser: i can do that too since i have multiple sim cards and SIP numbers to make automated calls Jan 10 14:21:26 lindi-: haven't tried it on FR but it works on an ARM7 dev board for me. Jan 10 14:22:04 In fact, in current situtation loading kernels to RAM with dfu-util is the easiest way i see. Jan 10 14:22:38 lindi-: are you setting up sime kind of regression-testing rig? Jan 10 14:22:42 some Jan 10 14:22:55 Copying a kernel into RAM: Jan 10 14:22:55 $ dfu-util -a 0 -R -D /path/to/uImage Jan 10 14:22:57 aha Jan 10 14:22:59 i gotta try this Jan 10 14:23:09 PaulFertser: so i first need to configure uboot to stay at the boot menu Jan 10 14:23:25 Kensan: planning to if i can manage Jan 10 14:23:30 Kensan: any help is appreciated Jan 10 14:23:33 hm, any problems known with the unofficial echo fix and shr? Jan 10 14:23:40 PaulFertser: 'Note: You cannot transfer more than 2MB of data into RAM using this method.' Jan 10 14:23:56 PaulFertser: '-rw-r--r-- 1 lindi lindi 2.4M Jan 8 17:13 uImage-moredrivers-GTA02_lindi-andy-tracking_b8b36e5ec3db71d5.bin' Jan 10 14:24:01 PaulFertser: so i have a problem Jan 10 14:24:38 lindi-: you can modify uboot to let it load larger kernels Jan 10 14:24:46 (at least that is what i read on the ML) Jan 10 14:25:10 Zorkman: it can already load them from SD here Jan 10 14:25:22 lindi-: you either have to hack on u-boot to add support for larger images. Or (much better!) you can dump u-boot altogether, prepare an initramfs you like (that kexec's kernel fetched over tftp or whatever) and append it to any working kernel. Flash Qi and the kernel and you're all set. Much more flexible. Jan 10 14:26:33 lindi-: remember that kexex works with zImage, not uImage. But i think u-boot header can be easily stripped from an uncompressed kernel especially. Jan 10 14:26:47 PaulFertser: kexec works here but is it the same? some hw might not be initialized the same Jan 10 14:27:25 PaulFertser: yes i have used kexec on freerunner ( see http://docs.openmoko.org/trac/ticket//2135 ) Jan 10 14:27:53 i applied the fix, made a phonecall, all worked fine, and now even after reboot contacts, messages, phonelog crashes Jan 10 14:28:50 somebody got the t-mobile g1 ? Jan 10 14:30:59 lindi-: the same with u-boot. I might initialize something not exactly as Qi. And Qi is the future for FR. u-boot is too unnecessary large and functional anyway. You surely can hack it to do what you want but are you sure it'd worth it? Jan 10 14:31:35 PaulFertser: i'm wondering why Qi is better :) Jan 10 14:31:53 PaulFertser: initramfs surely is but is that what you are going to do? Jan 10 14:32:09 lindi-: smaller, does only what's really necessary, easier to analize, fix bugs, more predictable (no environment) and so on. Jan 10 14:32:42 edje_object_part_swallow() seems to need a pointer as the last item, but i can't seem to find a way to set that pointer to another group in the edj file, any ideas? Jan 10 14:33:39 lindi-: u-boot goes the way legacy PC bioses go. And Qi is OM's coreboot. ;) Jan 10 14:36:05 PaulFertser: how should i have a boot menu that selects the kernel i want to boot with display and buttons? with initramfs and kexec? Jan 10 14:36:56 lindi-: yes, the most flexible way imho. Jan 10 14:37:27 hi Jan 10 14:38:21 PaulFertser: sure, i just didn't know that was the goal Jan 10 14:38:40 PaulFertser: that will surely increase boot time to have to boot two kernels? Jan 10 14:40:18 lindi-: you can prepare a very slimmed down kernel with only necessary facilities. In fact, angstrom project has already prepared something like that for the devices they support. I think it's more flexible. And boot time doesn't matter as much as hours of porting linux drivers to u-boot, imho. Jan 10 14:40:37 lindi-: well keeping Qi small practically means let's not do video initialization etc in Qi... Jan 10 14:40:59 all good arguments Jan 10 14:41:20 lindi-: so Qi is very simple but you can't do all that much with it. Jan 10 14:41:30 initramfs just wasn't mentioned in the previous Qi discussions or I have missed it Jan 10 14:43:33 It was, somebody's even going to implement a touchscreen-based menu. Jan 10 14:43:44 'grep -ri initramfs wiki.openmoko.org' from wiki is empty and 'grep initramfs irclogs/2008/freenode/#openmoko-cdevel.*' does not discuss Qi with it Jan 10 14:44:58 lindi-: it was discussed somewhere on MLs not long ago. Jan 10 15:11:48 very simple "test wlan, suspend, wait 60 secs, resume, test wlan ..." cycle is now running Jan 10 15:14:42 AntonTakk: add an edje object - then set the file TO that edje+group Jan 10 15:14:44 then swallow that obj Jan 10 15:14:45 :) Jan 10 15:19:13 now you can place bets on how long andy-tracking kernel will survive these constant resumes :) Jan 10 15:20:11 lindi-: is regulator stuff fixed in recent andy-tracking? I.e. does bluetooth work after resume? Jan 10 15:21:39 PaulFertser: i have not managed to get my bluetooth headset work so I have not been really using bluetooth for a while Jan 10 15:21:45 PaulFertser: i can add bluetooth tests later :) Jan 10 15:22:51 lindi-: it didn't appear as a device at all after resume. hcitool dev showed nothing and kernel oopses appeared in dmesg. Jan 10 15:29:20 PaulFertser: btw, is it possible to resume from suspend without having the display turn on? Jan 10 15:29:53 and also possible to make it not emit this "click" sound Jan 10 15:30:41 lindi-: have no idea, sorry. Jan 10 15:41:06 raster: thanks :) Jan 10 15:45:21 :) Jan 10 15:45:26 Good morning, Gentlemen Jan 10 15:58:57 abraxa_: hey, did you manage to solve your git problems? Jan 10 16:01:17 mrmoku! Jan 10 16:04:54 hehe, Dave :) Jan 10 16:05:34 http://scap.linuxtogo.org/files/27409ef78132499ca2da1667459796ad.png what's this? a modified omoney version? :p Jan 10 16:06:29 bumbl: script for pre-paid cards by yoyo Jan 10 16:06:38 ah ok Jan 10 16:06:50 but that would be a nice omoney feature Jan 10 16:06:52 too Jan 10 16:06:55 What, you didn't do it, bumble? Jan 10 16:07:06 bumbl: it work only for play provider (PL) Jan 10 16:07:20 i'll steal the idea yoyo Jan 10 16:07:28 nice would be illume "today" screen :P Jan 10 16:07:31 some gadgets Jan 10 16:07:36 piggybank! Jan 10 16:07:39 oink oink Jan 10 16:12:20 freesmartphone.org: 03mickey 07fso-monitord * r31b4d3d1a40b 10/src/obj.vala: add some logging functions Jan 10 16:12:36 freesmartphone.org: 03mok 07libframeworkd-glib * r447ad6711059 10/src/ogsmd/ (4 files in 2 dirs): Jan 10 16:12:36 freesmartphone.org: adjust to API change for SMS sent callback Jan 10 16:12:36 freesmartphone.org: Give the callback handler the new timestamp parameter Jan 10 16:12:38 freesmartphone.org: for the time when the message center received the message. Jan 10 16:12:40