**** BEGIN LOGGING AT Sun Jul 08 02:59:57 2007 Jul 08 03:42:09 http://bugs.openembedded.org/show_bug.cgi?id=2060 this bug has not been solved, any ideas about that Jul 08 07:27:26 * Bernardo is back. Jul 08 07:28:04 good morning Jul 08 07:51:20 03rwhitby 07org.oe.dev * rd09a058c... 10/ (3 files in 3 dirs): fsg3-kernel: Default kernel for Freecom FSG-3, with build defaults in fsgbe.conf to match vendor binary modules. Jul 08 09:04:43 * Bernardo is away: Ausente por agora. Jul 08 09:09:49 morning all Jul 08 09:13:02 hi all Jul 08 09:27:58 psokolovsky: hi Jul 08 09:28:19 psokolovsky: Do you happen to know what the current status of the hh.org htcuniversal kernels is? Jul 08 09:29:03 RP: Hi. Well, "very good'. What exactly do you mean? Jul 08 09:29:21 psokolovsky: I'm just wondering what the known problems areas are? Jul 08 09:29:41 psokolovsky: wifi seems "a little" unstable for example Jul 08 09:30:42 RP: well, its wiki page is still the best resource, http://handhelds.org/moin/moin.cgi/Universal Jul 08 09:31:00 RP: and yes, ACX100 wifi is not yet in perfect shape Jul 08 09:32:28 psokolovsky: ok, thanks. Is there much effort towards getting universal support into mainline? Jul 08 09:33:36 RP: well, we work on getting *every* machine's support into mainline, universal's support is on par with that ;-) Jul 08 09:34:00 psokolovsky: ok ;-) Jul 08 09:34:31 psokolovsky: I have a universal in front of me and I hate not using nearly mainline kernels ;-) Jul 08 09:35:42 RP: you can help to push our SoC chip drivers, that would be the great progress for both universal and bunch of other devices. Jul 08 09:36:06 psokolovsky: Does it have to use SoC or can it be switched to the device model? Jul 08 09:36:55 RP: SoC *is* the device model. Jul 08 09:37:18 psokolovsky: You just use platform devices now? Jul 08 09:38:00 psokolovsky: Sorry, its been a while since I looked closely at hh.org kernels ;-) Jul 08 09:38:50 psokolovsky: Have things like asic3 been attempted upstream? Jul 08 09:38:54 RP: Sure, how else it can be? We just organize them in such way that they track real architecture more faithfully, allow for cross-device reuse, and ultimately modular builds - all that to which mainline ARM-Linux not ready Jul 08 09:39:55 RP: yes. I submitted it for .22 timeframe, got bunch of things to cleanup, and that's where I am, and I don't have time to finish that for .23 timeframe. Jul 08 09:40:33 psokolovsky: Last time I Iooked at hh.org SoC was something very different (a reinvention of the device model) Jul 08 09:43:21 RP: http://lkml.org/lkml/2007/5/1/46 Jul 08 09:43:25 psokolovsky: The feedback on LKML for that patch doesn't look too bad Jul 08 09:43:29 psokolovsky: I was just reading that :) Jul 08 09:43:58 psokolovsky: I've seen much worse ;-) Jul 08 09:44:13 RP: yes, I'd say, it's ok, just need to finish acting on feedback, and push again, but as I said, I'm out of time for that now ;-( Jul 08 09:45:14 psokolovsky: I know that problem :/ Jul 08 09:45:40 psokolovsky: If I have some time I will look at it but I won't promise timescales. I doubt it will be this merge window... Jul 08 09:45:59 * RP needs to fixup the LED and backlight trees Jul 08 09:47:28 RP: All in all, we just currently discuss how we should handle such SoC drivers - 1) accept current ARM-Linux view that SoC is part of platform (== no modules); 2) use the current adhoc way to manage "child" devices, but use it consistently; 3) hack on LDD to teach it device dependencies. Jul 08 09:48:15 RP: well yes, I also don't think I'll get to this during summer, just hope to solve current q's/apply solutions consistently during this time. Jul 08 09:49:57 psokolovsky: Did you ever look at the locomo code, that had a similar "child" problem Jul 08 09:51:02 RP: I didn't actually looked deep enough, though I'm sure that many device would have this issue. So, can you build locomo as module? Jul 08 09:52:06 psokolovsky: The Kconfig says no, the code says yes Jul 08 09:53:34 psokolovsky: arch/arm/common/locomo.c - It creates its own bus type to handle the child devices Jul 08 09:54:34 RP: well, first part is actually enough for those who knows how (ARM) mainline treats something which doesn't fit into its simple picture ;-). as for code, must be function pointers and checking for NULLs? ;-) I'll have a look. Jul 08 09:55:07 RP: Ah, so *you* invent hacky bus types instead of using driver model? ;-) Jul 08 09:55:26 psokolovsky: No, its not hacky, creating a new bus is a perfectly legal part of the driver model Jul 08 09:55:38 psokolovsky: See the bus_register() call Jul 08 09:57:01 RP: ;-) That's exactly something for which older hh.org developers were larted when hey submitted original soc_bus implementation. So it's indeed funny ;-\ Jul 08 09:57:14 psokolovsky: I don't claim that code is perfect, just that perhaps it illustrates other ideas with potential Jul 08 09:57:37 psokolovsky: Do you have a pointer at that lart? Jul 08 09:58:31 I'll have a look at it now, but I guess, we put just much effort into cleaning up current "algorithm", exactly after mainline review suggested to get rid from non-standard buses. Jul 08 09:58:58 RP: not now, and don't think I could find it quickly, it was before me, 2004/2005 Jul 08 10:05:24 RP: Well, this is something about it, though I believe I saw even earlier attempts to submit it: http://lkml.org/lkml/2005/4/8/154 Jul 08 10:07:44 morning Jul 08 10:09:06 RP: Oh, and btw, maybe you can give a quick hint: do you have an idea how arch Jul 08 10:09:32 RP: arch's with pluggable CPU/APIC support deal with issue of assigning IRQs to them? Jul 08 10:09:37 psokolovsky: I do agree with the review there, that code is basically the platform device model, What locomo does is very different Jul 08 10:09:54 psokolovsky: I've never looked at that, sorry Jul 08 10:10:18 RP: well, so we use platform device model since that ;-) Jul 08 10:10:30 morning XorA Jul 08 10:11:09 RP: and anyway, main issue is not buses, but a dependencies among devices. that's the issue, and generic support for which mainline lacks. Jul 08 10:12:59 psokolovsky: I think I can see how the platform bus can be made to do that but I will fail to explain it on irc without code :/ Jul 08 10:15:32 RP: well, I see how that can be made on LDD level, but that won't be accepted mainline ever ;-). otherwise, we use pretty simple explicit way to deal with that on SoC driver level. And actually were considering if we fancy dropping it, but apparently, we *do* want to build single kernel for dozen devices, and don't want to compile all SoC driver statically. Jul 08 10:16:42 psokolovsky: You just need to play games with symbols to make that work properly without LDD changes ;-) Jul 08 10:18:18 RP: well, the problem is exactly that symbols don't work here ;-). Linux has pretty good dependency tracking for *drivers*, which is based on symbols. But *devices* is another domain, and you need to track pointers there. Jul 08 10:19:10 psokolovsky: I'm fairly sure there are games that can be played to make that work Jul 08 10:20:33 RP: ah, I bet yes, but we exactly strive for some generic and clean solution, even if in some niche (like SoC drivers, not whole LDD) Jul 08 10:21:03 psokolovsky: Sometimes you have to settle for some slight compromises to make mainline ;-) Jul 08 10:21:22 The trick is picking the compromises... Jul 08 10:22:40 RP: well yes, so I don't even consider hacking ldd. other approach is SoC drivers local, so won't give anyone probs, and no, we can't accept RMK's flat world view Jul 08 10:23:37 psokolovsky: I'll have a look at the code at some point, see if there isn't a way to work around the problems... Jul 08 10:24:28 RP: ok, thanks. I'll ping you when I'll clean up currently known issue in them. and I'll look at locomo for sure. Jul 08 10:24:53 psokolovsky: ok. I'll also let you know as/when I look at things in detail Jul 08 10:45:18 Laibsch: hi Jul 08 10:49:02 mrdata: Howd! Jul 08 10:49:06 Howdy! Jul 08 10:59:51 morning ! Jul 08 11:21:59 re Jul 08 11:38:07 hmm.. mtn: warning: attach node 2147499229 blocked by unversioned path 'packages/poboxserver/poboxserver/unicode.patch' - wtf? I don't have packages/poboxserver/poboxserver.... Jul 08 11:39:32 oops, figured this one. it's an `mtn mv` and it reports the other part :) Jul 08 11:41:42 okay, time to upgrade my production Z to the new layout :) Jul 08 12:04:13 polyonymous: is the building opie angstrom instructions still valid, I think I might try opie on my spare Zaurus Jul 08 12:06:23 XorA, I think yes. Jul 08 12:06:36 it's the result that differs now :) Jul 08 12:08:37 Is altboot not installed by default anymore? Jul 08 12:08:40 paypal has gone insane :-( Jul 08 12:08:41 I think it once was. Jul 08 12:08:47 I think it is. Jul 08 12:08:59 altboot is not by default on Angstrom Jul 08 12:09:13 I think it's good if itsn't, though. Jul 08 12:10:08 altboot breaks serial cable debugging :-) Jul 08 12:20:43 I'd also rather not have it in altbooted image :) Jul 08 12:21:16 altboot should be a choice, not by default Jul 08 12:21:51 * polyonymous agrees and chooses altboot. Jul 08 12:22:47 or we could have red pill/blue pill images :-) Jul 08 12:23:25 alboot + 7 partitions = WIN Jul 08 12:23:28 :)) Jul 08 12:23:43 summatusmentis, I have altboot + 1 partition and still enjoy it. Jul 08 12:23:54 lol Jul 08 12:23:55 I altboot into directories Jul 08 12:24:15 you can do that? Jul 08 12:24:22 I can :) Jul 08 12:24:39 can normal people? Jul 08 12:25:02 Well, I wouldn't recommend my way of doing it upstream, but I think it can be polished Jul 08 12:25:40 that'd be easy Jul 08 12:25:55 I could get rid of my partitions, and save space Jul 08 12:26:12 but now that I got qemuarm working I almost don't use altboot. Jul 08 12:27:03 how does qemuarm pertain to angstrom? Jul 08 12:27:47 well, it's not for use, it's for debugging. This is what I mostly use altboot for. Jul 08 12:27:55 I see Jul 08 12:28:37 * summatusmentis uses altboot for getting xfce,gpe,e17,opie and console all on one Z :-D Jul 08 12:29:13 I think I should think about my local hacks and maybe pass some stuff to coredump Jul 08 12:29:45 *nod* Jul 08 12:29:52 goodbye all Jul 08 12:30:19 bye Jul 08 12:34:46 hmm... ipkg install -force-reinstall doesn't seem to force reinstall... Jul 08 12:44:45 polyonymous: put the reinstall before install Jul 08 12:45:25 koen|inverness, ah. But strangely enough it looks as if it reinstalls. Not as if there was no -force-reinstall at all. Jul 08 12:45:56 koen|inverness: really? is that a requirement? Jul 08 12:46:19 with ipkg you have to put all the options in front or pray it doesn't blow up Jul 08 12:46:39 I think with ipkg you always should pray :) Jul 08 12:46:41 really? I didn't know that Jul 08 12:47:01 I've used -force-overwrite at the end Jul 08 12:47:25 and -recursive, as well as -force-removal-of-dependent-packages Jul 08 12:47:26 koen|inverness: Ive never put arguments before command and its always worked here Jul 08 12:47:36 well, like Winnie The Pooh would put it - you never know with ipkg. Jul 08 12:47:41 BTW, altboot is at least being built for spitz images, but not for collie images. And it is on my spitz installation. I don't recollect installling it explicitly which is why I asked. Jul 08 12:48:10 * Laibsch curses the big mess that OE (including angstrom) is becoming as of late Jul 08 12:48:53 RP: If you still around, I had a look at locomo.c. Well, that's exactly how our SoC drivers look like, which is no wonder, as I'm sure they were modelled either on Z ones, or on common ancestor. Except that we dropped usage of adhoc bus since that review and use platform_bus directly (which also means that we mostly dropped bus device IDs too btw). Jul 08 12:49:17 Laibsch, maybe you've installed it a while ago? Jul 08 12:50:47 polyonymous: No Jul 08 12:50:56 RP: But we have a bit wider problem, which apparently locomo doesn't expereince. Consider for example that standard PXA UDC uses EGPIO on some SoC chip for pull up control. Huh? That means that now UDC device depends on SoC device. If UDC is used before SoC driver is initialized, there's kaboom. Jul 08 12:51:21 Laibsch, well, I don't know, I also have it on the last spitz image I've installed. Jul 08 12:51:26 but I've built it a while ago Jul 08 12:51:45 polyonymous: Try 'MACHINE="spitz" bitbake -g opie-image; grep altboo task-depends.dot' and compare with 'MACHINE="collie" bitbake -g opie-image; grep altboo task-depends.dot' Jul 08 12:52:09 RP: that's exactly the problem I told about previously: in general case, *any* device may be dependent on *any* device. And we actually have cases like that UDC, and must deal with them. Jul 08 12:52:22 Laibsch, I have no reason to disbelieve you :) Jul 08 12:53:40 Laibsch: angstrom is not mess, and it's easy to keep it that way. for example, until Altboot pledges to support *all* devices, do not install it by default even for *single* one. And there will be order. Jul 08 12:54:35 angstrom-2007.1.conf:EXTRA_STUFF_append_spitz += "altboot" Jul 08 12:55:34 psokolovsky: Angstrom is becoming a mess. Just trying to clean up the shit left after the VERY important namechange on gmane Jul 08 12:56:04 So much wasted time lately. Jul 08 12:56:14 e6622d3a.. by Laibsch 2007-06-07: EXTRA_STUFF_append_spitz += "altboot" Jul 08 12:56:43 :)))))) Jul 08 12:57:26 Laibsch: I tell you a magic recipe how to keep it clean - do *not* allow for machine-specific discrepancies. either make it sure that it's indeed Ansgtrom's feature, or install in your own setup using ipkg. Jul 08 12:58:09 Q: "why is altboot installed on spitz" A: "because you commited that cset" Jul 08 12:58:23 that Jul 08 12:58:28 's funny indeed Jul 08 12:59:08 psokolovsky: I remember now. this was again a requirement from mr koen. Only make it default for supported devices. Jul 08 12:59:28 At the time I only knew that spitz was working. Jul 08 12:59:30 I don't think it should be default at all Jul 08 13:00:17 Laibsch: nice, just in case, here's suggestion from mr. pfalcon: add not universally supported software to generic distro very sparingly ;-) Jul 08 13:01:44 meh Jul 08 13:02:09 first days of "why isn't altboot in by default" and now "why is altboot in by default" Jul 08 13:02:25 it seems whatever the result people will complain Jul 08 13:02:38 can't one just add altboot via local.conf? Jul 08 13:02:47 yes Jul 08 13:02:58 in at least 5 different ways Jul 08 13:03:48 I'd suggest to go this way. Jul 08 13:05:13 feel free to remove it from angstrom.conf Jul 08 13:05:25 I stopped caring about 15 minutes ago Jul 08 13:21:51 I'd let Laibsch remove it. What if he would dislike me otherwise? :) Jul 08 13:27:15 koen|inverness: stop logging on while on holiday, its only upsetting Jul 08 13:27:42 for whom? :) Jul 08 13:29:07 polyonymous: I think Im the first OE person koen will bump into, I want to not get strangled with rage :-) Jul 08 13:29:30 :))) Jul 08 13:29:53 XorA: maybe you;ll get lucky and liam loses his key again ;) Jul 08 13:30:50 ~blame lrg_ Jul 08 13:30:53 * ibot blames lrg_ (and Canada) for all the evil in the world Jul 08 13:32:25 sweetness Jul 08 13:32:35 google just upgraded the imagery of edinburgh Jul 08 13:32:44 is it finally hires? Jul 08 13:32:56 http://maps.google.com/maps?q=55.949692+-3.171143+(Next+track+point)&z=16&t=h&om=1&iwloc=A Jul 08 13:33:56 that's the point where I realized I'm a fat bastard that wants a cookie instead of getting to the top of the hill Jul 08 13:34:38 wow edin isnt a blurry mess anymore Jul 08 13:35:15 koen|inverness: heh heh, you got a long way to go before you reach fat bastard claims :-) Jul 08 13:35:47 yeah Jul 08 13:36:02 and I actually lost some weight last month :) Jul 08 13:36:03 koen|inverness: its always best to climb that hill on the steep side after 10 pints Jul 08 13:36:13 koen|inverness: goes much quikcer that way Jul 08 13:36:48 * XorA is well into programmer size now Jul 08 13:37:50 what's a programmer size? :) Jul 08 13:37:55 does anyone know if linux iconv is thread safe? :> Jul 08 13:38:42 time to go watch some Dr Who Jul 08 13:38:47 bath time! Jul 08 13:38:58 * polyonymous thanks opie-reader :) Jul 08 13:50:09 Hi. I've been trying to build glibc on an arm machine, not exactly OE-related but wondering if anyone can help me? I don't seem to be adding the arm architecture specs in the right way, and am uncertain exactly what --build option to give to glibc configure. (cpu is detected as armv5tel) Jul 08 13:52:02 eekee: well, why do you think people use OE at all? to free their minds of such boring questions. someone makes glibc build well, and the whole community uses it now, working on real things, not reinvent the wheel. Jul 08 13:54:58 psokolovsky: reinventing the wheel is relative, if I learn this I'll be on my way without having to learn a bunch of other things. :) Jul 08 13:56:32 Actually if monotone would compile I wouldn't be asking, I'd be looking at OE & distro sources, it's just that compiling monotone feels like I'm getting even further sidetracked. maybe I should see if a binary build of monotone will work... Jul 08 13:56:33 eekee: sure, glad for you. just don't be surprised if you won't receive the answer here, I explained above why. people here don't keep that stuff in mind, they teach OE to do that, and then OE remembers all the gory details. Jul 08 13:57:54 ok, fair enough Jul 08 14:37:38 hi Jul 08 15:57:13 woglinde: hi Jul 08 15:57:17 jo Jul 08 15:59:48 hi koen Jul 08 15:59:53 still rainy? Jul 08 16:03:44 /wo 39 Jul 08 16:03:46 oops Jul 08 16:28:00 * * OE Bug 2615 has been created by gfellerpatrik(AT)gmx.net Jul 08 16:28:02 * * unable to build libpng-1.2.12-r5: task do_install failed Jul 08 16:28:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2615 Jul 08 16:45:33 hi ! Jul 08 16:45:47 eekee: still there ? Jul 08 16:46:01 yep Jul 08 16:46:06 hi ^^; Jul 08 16:46:52 eekee: I have built monotone 0.32 recently. What was your problem whenbuilding ? Jul 08 16:48:04 I was trying to build 0.35 with boost 1_34_0. Pasting the errors Jul 08 16:48:10 numeric_vocab.hh:25: error: 'CHAR_BIT' was not declared in this scope Jul 08 16:48:10 numeric_vocab.hh:25: error: template argument 1 is invalid Jul 08 16:50:38 eekee this know Jul 08 16:50:53 eekee try the latest packages from the monotone side Jul 08 16:51:05 eekee: what kind of system do you have - is it debian - based ? E.g. I compiled monotone on an elive system, which is debian based, so I just installed the boost libraries by apt-get Jul 08 16:52:06 eekee: Just build 0.32, that worked for me. Jul 08 16:52:25 y Jul 08 16:52:27 monotone has 0.35 debs which wirks Jul 08 16:52:30 I have described what I did in detail here: http://www.stoefer.com/index.php?topic=projects&project=netbook_oe&part=2 Jul 08 16:52:32 aeh Jul 08 16:52:37 works Jul 08 16:52:39 he mickeyl Jul 08 16:52:49 yo Jul 08 16:53:32 woglinde: that is the latest Jul 08 16:54:35 jan_js: Source Mage GNU/Linux, it tries to stick as close as possible to what the apckage maintainer supply, no unnecessary patches etc Jul 08 16:55:03 jan_js: thanks for the sugestion though, and for the link ^^ Jul 08 16:56:00 eekee: hmm.. never heard about it. Yes, hope the description will help you ! Jul 08 16:56:12 ty ^_^ Jul 08 17:10:18 polyonymous: kernel/qgfxw100_qws.cpp:1253: error: 'clipbounds' was not declared in this scope Jul 08 17:10:33 polyonymous: seent that in qte build? Jul 08 17:10:45 never. Jul 08 17:11:16 I'm willing to reproduce it if it's possible at all ;) Jul 08 17:12:38 polyonymous: thats building for c7x0 so I assume thats the only machine that builds the w100 accellerationh Jul 08 17:13:10 MACHINE=c7x0 bitbake qte-mt ? Jul 08 17:13:49 polyonymous: I was that should do it Jul 08 17:13:56 lemme try Jul 08 17:14:42 arg all panic, peeron is down Jul 08 17:18:08 reproduced :) Jul 08 17:19:32 polyonymous: not my insanity then :-) Jul 08 17:19:44 no, not it. Will you fix it or should I look into it/ Jul 08 17:21:48 polyonymous: if you dont get round to it I will eventually Jul 08 17:22:02 polyonymous: but at moment I am building legos Jul 08 17:24:13 ok, I may, but I have no c7x0 :) Jul 08 17:26:01 damn them lego torches Jul 08 17:30:43 03polyonymous 07org.oe.dev * r35f0f01e... 10/ (4 files in 3 dirs): opie-irc: utf8-ize Jul 08 18:06:49 polyonymous: also went upstream way sure? Jul 08 18:17:26 psokolovsky, what? :) Jul 08 18:17:37 psokolovsky, are you talking about opie-irc? Jul 08 18:17:56 polyonymous: yes Jul 08 18:18:47 psokolovsky, not yet, (a) Darth_Wader says there's another issue left and (b) I wouldn't say these patches are perfect, but I sure would let anyone taking care of opie-irc decide on it. Jul 08 18:19:02 psokolovsky, but I still need to look into russian channel topic problem Jul 08 18:19:52 ok Jul 08 18:28:27 XorA, I think I'll push the build fix in a few minutes Jul 08 18:28:41 it's a gcc4 thing Jul 08 18:29:50 polyonymous: cheers, Im still building legos :-) Jul 08 18:30:05 what legos? :) Jul 08 18:30:24 http://www.peeron.com/inv/sets/6380-1?showpic=7773 Jul 08 18:30:36 ah, thought so Jul 08 18:31:00 * * OE Bug 2616 has been created by gfellerpatrik(AT)gmx.net Jul 08 18:31:02 * * package libxcb-0.9.93-r2: task Jul 08 18:31:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2616 Jul 08 18:33:05 XorA: I have that same treatment center :) Jul 08 18:35:03 koen|inverness: I am building it from spares Jul 08 18:36:24 * CoreDump|bbl is having his monthly "backup day" Jul 08 18:41:29 XorA, pushed (I don't rely on cia anymore :)) Jul 08 18:42:23 polyonymous: cool, pulling Jul 08 18:42:49 now we'll see if you finish building before cia speaks up :) Jul 08 18:51:53 http://biagi.miniville.fr Jul 08 18:52:01 * * OE Bug 2617 has been created by utx(AT)penguin.cz Jul 08 18:52:03 * * sound mixer misconfiguration Jul 08 18:52:05 * * http://bugs.openembedded.org/show_bug.cgi?id=2617 Jul 08 18:58:01 03mickeyl 07org.oe.dev * r94ba8615... 10/ (10 files in 4 dirs): Jul 08 18:58:01 add alsa-states, containing the machine specific alsa states. Jul 08 18:58:01 TODO: Find out how to interact with the alsa-state package regarding the default state Jul 08 18:58:06 03mickeyl 07org.oe.dev * r3b04f9d3... 10/ (6 files in 3 dirs): poppler: add recent cvs version since 0.5.9 release doesn't build Jul 08 18:58:12 03polyonymous 07org.oe.dev * rf6a57e1a... 10/ (3 files in 3 dirs): qte 2.3.10: fixed build for c7x0 with gcc 4 Jul 08 18:58:52 due to internal load-levelling issues our CIAbot now comes with more latency Jul 08 18:59:04 we will improve that again -- eventually Jul 08 19:00:11 * polyonymous pats poor cia Jul 08 19:01:59 XorA: http://linuxtogo.org/~koen/Edinburgh-walk.jpg Jul 08 19:03:03 polyonymous: aha, so you can confirm that c7x0-w100-accel.patch works? Jul 08 19:03:18 koen|inverness: did you track your movement? Jul 08 19:03:33 Jin^eLD: yes nokia n800 + gps + maemomapper Jul 08 19:03:42 (then gpsbabel'ed to kml) Jul 08 19:03:44 that's damn cool! :) Jul 08 19:04:11 psokolovsky, no, I can confirm it builds, ask XorA whether it works :) Jul 08 19:04:11 google added hi-res imagery for edinburgh today :) Jul 08 19:04:11 koen|inverness: so, that's standard googleearth? Jul 08 19:04:16 he spotted the build problem Jul 08 19:04:20 koen|inverness: you apparently got beamed to the spot on top right ;) Jul 08 19:04:20 psokolovsky: yes Jul 08 19:06:20 psokolovsky: compiles here Jul 08 19:07:03 XorA, polyonymous: would be nice to know it works, and them merge it in such way that it was always compiled into qte. Jul 08 19:07:31 psokolovsky: I will test it when I get image built Jul 08 19:07:37 I think it doesn't apply to spitz? Jul 08 19:07:40 nice Jul 08 19:08:01 psokolovsky: opie is lots of c++ give my poor machine a chance Jul 08 19:08:02 polyonymous: it's machine-specific patch now Jul 08 19:08:34 psokolovsky, yes, but would spitz benefit from it? I actually have no idea about the patch Jul 08 19:08:46 spitz has no w100 Jul 08 19:08:48 only c7x0 has w100 Jul 08 19:08:55 (and hx4700-ish) Jul 08 19:08:57 polyonymous: QA and maintenance would benefit. Jul 08 19:09:13 koen|inverness: cool map Jul 08 19:09:34 psokolovsky, ah ok. Then just add it to the top-level srcuri? Jul 08 19:10:21 polyonymous: yes, if it's nice patch which doesn't disable non-w100 use ;-). that's exactly sth to check. Jul 08 19:11:24 psokolovsky, I don't mind moving it around the .bb, nobody needs a written permission from me, even thought is' opie(tmp)-related :) Jul 08 19:12:10 polyonymous: well, I hint that someone should review what exactly that patch does, and how w100 usage is configured in qte. Jul 08 19:13:23 psokolovsky, I got it, but it's not really my realm of understanding. Jul 08 19:13:42 polyonymous: you must be kidding ;-) Jul 08 19:14:00 fwiw, the author of that patch uses X11 now (sirfred, xserver-kdrive-imageon) Jul 08 19:14:16 psokolovsky, I have never ever looked into it. Jul 08 19:14:40 i'm afraid the w100 qte patch has never made it into production state Jul 08 19:14:53 feel free to play with it, but it may eat your children Jul 08 19:17:30 mickeyl: that's what I want to know, if it would make more sense to drop it, taking into account opie's orientation towards low-end devices... Jul 08 19:17:50 Well, let's wait XorA's experience Jul 08 19:19:48 ya Jul 08 19:28:59 mickeyl: http://rafb.net/p/3HFgkb30.html <- ecore-native build failure Jul 08 19:29:41 koen|inverness: bummer, i can not reproduce that Jul 08 19:29:47 which machine/distro ? Jul 08 19:30:07 ecore-*native*, so machine and distro don't matter Jul 08 19:30:17 ah Jul 08 19:30:19 well Jul 08 19:31:10 fresh build? Jul 08 19:31:14 DISTRO=angstrom-2007.1 and any machine in ep93xx h6300 fic-gta01 a780 a1200 e680 rokr-e2 at91sam9263ek qemuarm h2200 h4000 nslu2le ixp4xxle nokia770 omap5912osk poodle tosa h5000 magician htcuniversal htctornado hx2000 hx4700 c7x0 spitz akita netbook-pro progear guinness qemux86 mx31ads nokia800 efika nslu2be ixp4xxbe collie h3600 h3800 simpad htcwallaby Jul 08 19:31:15 sure you don't have any leftovers? Jul 08 19:31:19 yes, fresh build Jul 08 19:31:44 strange. ok, i'll investigate more. ecore-native should not use software-x11 if it can't find it. Jul 08 19:31:56 yes, i know the simple way would be to just enable it Jul 08 19:31:58 and I don't have evas headers on my host Jul 08 19:32:02 but i want to know why it doesn't work :D Jul 08 19:32:14 koen|inverness: nod I'll build on our buildhost tomorrow Jul 08 19:32:19 (which you probably do have) Jul 08 19:32:20 that one is clean as well Jul 08 19:32:23 (header-wise) Jul 08 19:32:25 yep Jul 08 19:34:16 | checking for .preinit_array/.init_array/.fini_array support... no Jul 08 19:34:17 | configure: error: Need linker with .init_array/.fini_array support. Jul 08 19:34:17 | FATAL: oe_runconf failed Jul 08 19:34:35 from glibc-intermediate for ANgstrom/omap5912osk Jul 08 19:34:41 Crofton: 'unset MACHINE ; bitbake binutils-cross -c rebuild' Jul 08 19:34:44 any thoughts? Jul 08 19:35:09 is this temporary? Jul 08 19:35:20 don't set MACHINE in env :) Jul 08 19:35:28 hmm Jul 08 19:35:29 * koen|inverness feels a deja vu Jul 08 19:35:43 where should zI set machine? Jul 08 19:36:01 in local.conf or auto.conf Jul 08 19:36:08 I've been setting it in the env so I do not have to gard code it, so Ican esaily switch between efika and OSK Jul 08 19:36:43 you can set it in env after building glibc Jul 08 19:37:08 grr Jul 08 19:37:15 This is not a "feature" Jul 08 19:39:13 running bitbake binutils-cross without machine set flunks sanity check Jul 08 19:39:25 right Jul 08 19:39:32 so set it in local.conf or auto.conf Jul 08 19:39:37 yep Jul 08 19:39:51 mainly pointing that out for people reading the log :) Jul 08 19:42:29 good thing you took a train to Scotland .. Jul 08 19:44:00 * * OE Bug 2618 has been created by utx(AT)penguin.cz Jul 08 19:44:02 * * Clean spitz.c ALSA driver Jul 08 19:44:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2618 Jul 08 19:46:24 good morning Jul 08 19:46:34 whois ? Jul 08 19:50:36 RP: ping Jul 08 19:51:24 koen|inverness, Crofton: binutils-cross must have been fixed regarding MACHINE in env long ago. And I did check it was fixed. Jul 08 19:53:02 ~botmail for RP: do we have a known problem with shlib package dependency granularity? say i have a bb that emits a library and some testing program. the testing program depends on additional libraries, however even the library package gets all of the testing-program dependencies (which is wrong). Jul 08 19:57:59 * * OE Bug 2618 has been RESOLVED (WONTFIX) by dp(AT)xora.org.uk Jul 08 19:58:01 * *  Clean spitz.c ALSA driver Jul 08 19:58:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2618 Jul 08 20:14:07 03mickeyl 07org.oe.dev * red388eba... 10/ (1 packages/efl1/epdf_cvs.bb): add epdf, the glue between libpoppler and the EFL Jul 08 20:14:11 03mickeyl 07org.oe.dev * r95d5e6ff... 10/ (3 files in 2 dirs): add epsilon_cvs to get the shiny new epdf thumbnailer plugin Jul 08 20:39:40 XorA: suse employee (stanilav brabek) Jul 08 20:54:30 night all Jul 08 21:19:07 hi all, I want to create a propper .bb file for my programm. this is my first draft (shortlink.org/9161). the programm needs evdev and uinput. I was searching some .bb files and it seems to be a common way to RRECOMMENDS thouse modules and DEPENDS = "virtual/kernel". is this still a good idea? Jul 08 21:21:56 03polyonymous 07org.oe.dev * rb6e5034a... 10/ (3 files in 3 dirs): opie-irc: more fixes for utf-8 Jul 08 21:27:09 * wick7xx is away: afk a few hours Jul 08 21:50:31 mr_nice: hardly. userspace applications should not depend on kernel modules. Jul 08 21:51:32 psokolovsky: in general or in oe? Jul 08 21:51:36 but they can RRECOMMEND them Jul 08 21:51:52 mr_nice: in general at all Jul 08 21:53:26 CoreDump: sure, why not. they can recommend glibc for example too. they can even rdepend on it. or on specific version of it. or on specific kernel version. and many other things they can do. but why? Jul 08 21:53:43 anyway... Jul 08 21:53:49 psokolovsky: hm, i have no idea how else i could have implemented the program. it reads evdev input blocks and parse it and makes a uinput device to submit the changed evinput Jul 08 21:54:06 like it or not, there are indeed userland apps out there which require certain kernel features Jul 08 21:55:14 mr_nice: so you would need to check that that evdev in uinput is there during runtime, and five out error if they are not. Jul 08 21:56:33 CoreDump: features, yes. but putting into userland app package's a dependency on specific kernel module is inherently flaky IMHO. Jul 08 21:56:56 correct, that's what RECOMMEND is for Jul 08 21:57:17 "install if available, but don't bitch if it isn't" Jul 08 21:57:34 psokolovsky: hm, good idea. that is something I real have forgotten to check. Jul 08 21:58:11 ok so i will RRECOMMEND on the modules and put some checks into the program Jul 08 21:58:56 mr_nice: that would be a sane approach Jul 08 21:59:05 thx :) Jul 08 22:28:26 is there a list of mentors for angstrom machines somewhere? Jul 08 22:31:16 hm look at Maintainers file Jul 08 22:31:23 only source for now Jul 08 22:32:24 woglinde, where can I find that? Jul 08 22:33:42 in your oe repos? Jul 08 22:33:55 org.openembedded.dev/MAINTAINERS Jul 08 22:35:24 nite Jul 08 22:38:12 * * OE Bug 2564 has been RESOLVED (FIXED) by Jul 08 22:38:14 * *  QT/E fails to build due to lack of asm/page.h Jul 08 22:38:16 * * http://bugs.openembedded.org/show_bug.cgi?id=2564 Jul 08 23:19:58 nite Jul 08 23:31:43 03polyonymous 07org.oe.dev * r3f2e7520... 10/ (3 files in 3 dirs): poboxserver: adjust to its new home in /usr Jul 08 23:31:53 03polyonymous 07org.oe.dev * rc59f1131... 10/ (1 packages/nunome/nunome_1.0.2.bb): nunome: move to the new opie location Jul 08 23:37:00 * * OE Bug 2602 has been RESOLVED (INVALID) by lukas.gorris(AT)gmx.de Jul 08 23:37:02 * *  zaurus c1000 akita touchscreen broken Jul 08 23:37:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2602 Jul 08 23:40:53 polyonymous: Any changes except the move to /usr? Jul 08 23:41:11 Laibsch, no, I think not. Jul 08 23:41:18 but I've tested it as much as I could :) Jul 08 23:42:00 * * OE Bug 2582 has been RESOLVED (FIXED) by lukas.gorris(AT)gmx.de Jul 08 23:42:02 * *  sd-card is not correctly recognized by kernel Jul 08 23:42:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2582 Jul 08 23:42:30 Laibsch, but as it always goes with this brilliant software - without adjusting to /usr it would kill opie :) Jul 08 23:46:59 * * OE Bug 2619 has been created by lukas.gorris(AT)gmx.de Jul 08 23:47:02 * * Meta-Bug for SL-C1000 (akita) Jul 08 23:47:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2619 Jul 08 23:47:59 * * OE Bug 2407 has been RESOLVED (FIXED) by lukas.gorris(AT)gmx.de Jul 08 23:48:02 * *  build of sysvinit fails in do_install Jul 08 23:48:04 * * http://bugs.openembedded.org/show_bug.cgi?id=2407 Jul 09 00:13:47 morning Jul 09 00:15:44 or night Jul 09 01:06:56 could somebody tell me what the package gpe-windowlist is supposed to do? Jul 09 01:07:17 is it like a list of windows for the panel? Jul 09 02:57:25 03tmbinc 07org.oe.dreambox * re646a959... 10/ (1 conf/machine/dm7025.conf): conf/machine/dm7025.conf: don't mess up rootfs while creating image **** ENDING LOGGING AT Mon Jul 09 02:59:59 2007