**** BEGIN LOGGING AT Fri Dec 16 02:59:56 2011 Dec 16 05:43:04 lilstevie: hey, is the adbd stuff open-source (client and/or server side)? Dec 16 05:43:41 btw for writing your init.d scripts you might want to look at update-metainit or just upstart jobs Dec 16 05:44:29 http://paste.debian.net/149355/ before; http://paste.debian.net/149356/ after Dec 16 05:45:09 seriously Dec 16 05:45:31 "after" is upstart job Dec 16 05:45:39 I am sick of people saying "you can do this" or "you can do that" or "you should do this/that" instead of telling me, do it, then contribute Dec 16 05:45:55 :) Dec 16 05:46:03 so-rry Dec 16 05:46:22 This is what happens when you pick a project that weenie end users are attracted to Dec 16 05:46:39 if you were writing e.g. a replacement DNS server... Dec 16 05:46:42 :-) Dec 16 05:49:30 twb: But the thing is, things are so early in development such that such things are not really a priority right now. We are starting to talk about proper Ubuntu integration and getting packages into the archive, but we don't really have the basics sorted yet. Dec 16 05:49:58 So whatever works now for testing that is quick and convenient is used. Dec 16 05:50:05 Sure thing Dec 16 05:50:18 I'm just pointing out stuff as its in front of me Dec 16 05:50:46 Back to my question: is adbd free software? Dec 16 05:51:46 adbd is part of AOSP yes Dec 16 05:51:47 I dunno. Dec 16 05:51:54 ah ok Dec 16 05:52:34 Goody Dec 16 05:52:55 and yeah, everything is quick Dec 16 05:53:18 and hacky Dec 16 05:53:49 I lived through some of that over on #debian-eeepc too, back when x86 EEEs needed extra magic for various things Dec 16 05:54:12 everything is magic here Dec 16 05:54:18 Deep magic Dec 16 05:54:45 * XorA predicts he will have time to look at private projects in about 2042 Dec 16 05:55:01 XorA: that when the gypsy said you'd die? Dec 16 05:55:13 when Im 65 and I retire :-) Dec 16 05:55:30 heh Dec 16 05:56:21 by then things like tegra2 will be as retro as my zx81 Dec 16 05:57:22 and I will still be trying to find this damn uart Dec 16 05:58:34 Well, think about the oldest chip design you're actively using Dec 16 05:58:41 * XorA unfortuneately doesnt yet have a tegra NDA so cant tell you where it is Dec 16 05:58:54 There's a purple sparc pizza box in front of me, that's probably 25 years old Dec 16 05:59:11 So in 2035ish I could expect to be babysitting tegra2s Dec 16 05:59:32 XorA: uh, you know what NDA means, right? Dec 16 05:59:56 twb: yes, why? Dec 16 06:00:08 If you signed an NDA you *wouldn't* be able to tell Dec 16 06:00:59 XorA: and it is the pinout on the board I am trying to find it with, changes vendor to vendor Dec 16 06:01:07 NI adam has a really easily accessable one Dec 16 06:01:31 twb: they are normally not that secretive, but do give nice easy access to docs Dec 16 06:11:43 anyone have any idea why wireless would set the default route to end in .0 instead of .1. result i can access the world, but the local wireless network cannot be seen (ping, nslookup) other than the access point Dec 16 06:36:07 MrCurious_: Because your DHCP server is handing out bogus information? Dec 16 06:40:18 working with a bud on it, and we just came to that conclusion as no machines on the wireless can ping any other machine on wireless Dec 16 06:58:39 The NDAs I've signed basically say that you're not allowed to tell anybody anything Dec 16 07:28:36 twb: yeah, like even "you may not mention this NDA" Dec 16 07:28:46 Yeah Dec 16 07:28:52 Although that's more defamation stuff Dec 16 08:44:43 Guess what happens without keyboard repeat Dec 16 08:44:53 Can't hold ^C to kill a tight sh loop :-/ Dec 16 10:10:24 lilstevie: fyi, I think some of the funkiness with the onboard keyboard and holding down the mod keys, is specific to holding down the caps lock key, which I have set to be a control key Dec 16 10:10:50 I *think* it's harder to reproduce some of the problems when using the key originally intended to be control Dec 16 10:28:31 Just out of curiosity, if you guys don't mind me asking, how's the ARM version going? Dec 16 10:29:28 pretty well, the switch to hardfloat has happpened (now it needs to provide stable until feature freeze) so we might drop support for armel Dec 16 10:41:37 ogra_: thats a bit much isn't it :p Dec 16 10:44:16 lilstevie, hmm ? Dec 16 10:44:37 whats a bit much ? Dec 16 10:44:56 dropping armel Dec 16 10:45:10 the plan is to only support one arm flavour ... and that will likely be armhf by the looks of it Dec 16 10:45:20 I see Dec 16 10:45:33 12.04 is an LTS isn't it Dec 16 10:45:38 we wont wipe the armel archive :) Dec 16 10:45:45 but we wont support it at all Dec 16 10:45:57 ah ok Dec 16 10:46:03 12.04 is LTS for all arches but arm Dec 16 10:46:15 just wondering cause you know L4T drivers and the likes Dec 16 10:46:19 ah ok Dec 16 10:46:28 is arm going to end up with an LTS any time soon? Dec 16 10:46:30 it might *become* LTS for arm server if we recieve hardware in time Dec 16 10:46:39 but thats server only Dec 16 10:46:42 ok Dec 16 10:47:17 desktop/client will be non LTS .... as well as all the images will Dec 16 10:47:41 lilstevie: nvidia can if asked nicely provide armhf drivers if they find your usecase interesting enough Dec 16 10:47:53 (though its not like you wouldnt get userspace updates fro free from x86 uploads indeed ;) ) Dec 16 10:48:23 xranby: I know that, just that isn't their key build yet Dec 16 10:48:24 :) Dec 16 11:03:37 What is the status of ARMv6 support at the moment? Dec 16 11:03:52 I know it will not be officially supported, but you always mention that you'll not *remove* the support Dec 16 11:04:12 WaltherFI: not supported by ubuntu since ubuntu target armv7 Dec 16 11:04:13 so If I understand that correctly, there is some sort of compatibility? Dec 16 11:04:22 WaltherFI: debian armel work Dec 16 11:05:14 WaltherFI: there exist some armv6 chips with thumb2 support that can in theory run some of the programs Dec 16 11:05:16 WaltherFI, there is no compatibility and armv6 support isnt planned at all Dec 16 11:05:34 with armhf we even move further away from v6 Dec 16 11:06:23 since the hardfloat capabilities fully rely on v7 hardware Dec 16 11:27:23 So no love for ARM11 that is. Dec 16 11:28:17 WaltherFI: there has been no love for arm11 since karmic Dec 16 11:28:32 no ubuntu love for the raspberryPI ;( Dec 16 11:29:29 RaTTuS|BIG: Sad. Even more sad now that they will release the additional I/O board --> robots! Dec 16 11:32:09 it's a pity as I run ubuntu on everythign else [well apart from the obigitary windows machines for work] but we'll cope... Dec 16 11:37:22 WaltherFI, well, i am telling you the same thing since 3 months, asking the same question over and over will not change reality Dec 16 11:39:19 RaTTuS|BIG, if someone provides the resources (build machines, archive space on the servers etc cost a lot of money) and forms a team (a few people that constantly and reliable take responsibility ) to do a v6 port, nobody would stop you doing a community v6 archive ... Dec 16 11:41:24 ubuntu did not support v6 since 2 years, that we supported it at all was not by plan but due to that fact that we started off the debian port initially, there was never a plan to keep pre-v7 support Dec 16 11:43:15 (and this was communicated at all UDSes and in other public media, its not a secret or anything and never has been) Dec 16 11:44:39 if you want to support old arm specs, simply use debian, its not like debian is *that* much different from ubuntu Dec 16 11:49:12 ogra_: I'm sorry if I sound repetitive - I just wanted to know what is the level of compatibility, as there have been mentions about the support being 'dropped but not removed' Dec 16 11:49:30 armel wont be removed Dec 16 11:49:42 but nothing changed beyond that Dec 16 11:50:53 And what is the status of armel at the moment? Do even the very essential packages build at least? Dec 16 11:51:17 armel wasnt touched much this release, its the same as in oneiric Dec 16 11:51:24 ogra_ - yeah it's something I may do .... anyway let see what my work load is in the new year ;-p Dec 16 11:51:33 our focus was on building armhf Dec 16 11:52:51 (which took a team of ~five people working full time constantly on it for a few months ... just to give you an impression how much work it is to change a port ... if someone wants to do v6 that will be similar) Dec 16 11:52:55 So - does oneiric work then with armv6? Is it installable, that is Dec 16 11:53:30 WaltherFI, you ask the same thing all the time ... ubuntu didnt support v6 since two years and wont support v6 in the future Dec 16 11:54:16 if someone wants to do a community port and will pay the money for the resources too, a v6 port could happen but i wouldnt count on it Dec 16 11:55:53 ogra_: I understand that there is no support, but I'm trying to ask about the level of the armel as it has been left; i.e. even though the support has been removed, what is the level of the last version you have touched Dec 16 11:56:05 I'm sorry if I'm beginning to annoy you, sincerely Dec 16 11:56:09 honestly* Dec 16 11:56:25 (instead of paying money to have resources in the ubuntu infrastructure you can indeed run your own datacenter somewhere else, doesnt need to be money, but you will need teh resources) Dec 16 11:56:53 I am just trying to figure out whether the said community project would have to be started from the beginning or do you have something done from the times it was supported Dec 16 11:56:57 WaltherFI, as i said, armel didnt (and likely wont ever) change ... its v7 only Dec 16 11:57:48 Ah, I'm sorry, I've somehow read your responses as there would have been support at some point, back the said two years ago or so Dec 16 11:57:52 for starting such a port you would have to take armel, switch compiler defaults and rebuild the whole archive in the right order Dec 16 11:58:10 and then fix all build failures (which can be 100s) Dec 16 11:58:50 there was support for v6 three years ago when we pulled the arm port from debian Dec 16 11:58:56 Sure, I understand that porting it would require a lot of resources Dec 16 11:59:05 that support was never meant to stay and was dropped two yeras ago Dec 16 11:59:12 and I appreciate all the work you've done for the armv7 support Dec 16 11:59:22 it was just out of technical reasons that v6 was there at all Dec 16 11:59:43 since we didnt want to directly switch to v7 when pulling from debian Dec 16 12:00:13 Ah, so there is something that has been done already - is the work you did for building armv6 back then saved somewhere or is it deleted? Dec 16 12:00:54 it happened six releases ago, its gone Dec 16 12:01:22 Thank you, that was the information I was looking for. Sorry for this Dec 16 12:01:25 jaunty was supporting v5, karmic was then switched to v6, with lucid we didnt the full switch to v7 Dec 16 12:01:31 old release are somewhere on the web Dec 16 12:01:38 s/didnt/did/ Dec 16 12:01:53 right, there is the old-releases.ubuntu.com archive Dec 16 12:02:02 that should have jaunty and karmic Dec 16 12:02:48 Nice! I'll take a look at that then Dec 16 12:03:09 that wont gain you anything if you want to be up to date though Dec 16 12:03:35 just taking the existing armel port and re-rolling it somewhere is likely less work Dec 16 12:03:54 all the SW in jaunty is long outdated Dec 16 12:04:59 mmhmm Dec 16 15:55:49 im trying to get ubuntu onto netbook with800 mhz telechip processor Dec 16 22:44:13 infinity, do you think this would suffice to force the build ith gcc 4.5 ? http://paste.ubuntu.com/772746 Dec 16 22:44:23 I am now trying locally with a cross build Dec 16 22:44:44 the issue with non-booting ac100 kernel goes away if built with 4.5 vs 4.6 Dec 16 22:45:07 it worked for me because I had not dist-upgraded yet, whereas the buildds had 4.6 already Dec 16 23:07:48 janimo: *blink* Dec 16 23:08:03 janimo: What do you mean "not dist-upgraded yet"? gcc-4.6 was the default in oneiric too. Dec 16 23:08:25 janimo: And if there are bugs being exposed by 4.6, we should fix them, not paper over them. Dec 16 23:08:33 infinity, not dist-upgraded from 4.6.1 to 4.6.2 gcc-arm Dec 16 23:08:42 The 4.6.1 build worked too Dec 16 23:09:02 I just tried with 4.5 now. So yes, the problem can be narrowed down maybe Dec 16 23:09:30 janimo: Trying to narrow it down would be nice... Dec 16 23:09:44 infinity, we should fix them indeed, but with the pace bugs get tracked down and fixed in gcc we may not get new kernels for a while Dec 16 23:10:08 infinity, the fact that the machine shows the toshiba logo and does not boot makes it hard to debug Dec 16 23:10:25 It's less likely to be a gcc bug and more likely a bug in your source tree, to be honest. Dec 16 23:10:46 Forcing 4.5 is fine for now, though, since it's only a universe kernel. :/ Dec 16 23:11:39 infinity, you mean a bug which building with 4.5 manages to be avoided? Dec 16 23:12:33 We had a similar case with QT earlier this year. A volatile added made it work when built with 4.5 too but would only work with 4.4 otherwise Dec 16 23:13:06 janimo: There were lots of sketchy bugs fixed in the kernel over the last couple of years to make it happier with 4.6. Dec 16 23:13:24 janimo: Non-mainline trees often repeat the mistakes already learned in mainline, sadly. Dec 16 23:14:21 (In fact, didn't we fix some ARM-specific kernel bugs just last cycle that people thought were GCC-related?) Dec 16 23:14:48 Something about misaligned structs or some such. Dec 16 23:15:03 Old person memory not working well here. :P Dec 16 23:15:20 I am not familiar with kernel work from last cycle Dec 16 23:15:48 Maybe I'm confusing the kernel with something else. Dec 16 23:15:49 I'll see if there's some new warnings in the 4.6 build vs the 4.5 one Dec 16 23:16:17 It's my vacation, I don't have to be smart. Dec 16 23:16:37 But if you want to upload with a build-dep on 4.5 for now, go ahead. Just remember you've done so, cause we should fix it. Dec 16 23:16:50 I honestly wish gcc stopped being upgraded besides bugfixes for a while and let apps go ahead Dec 16 23:16:59 (And test your fix in a clean chroot to make sure it does what you think it does) Dec 16 23:17:12 infinity, yeah, I'd remember, it is not many packages that I change to explicitly use another gcc :) Dec 16 23:17:42 infinity, your memory is fine there were some weird packed structs in ehci code in the kernel that did not work with 4.6 Dec 16 23:17:46 infinity, does debuild work with cross chroot? Dec 16 23:18:09 janimo: I dunno. I don't cross compile anything. Dec 16 23:19:01 jcrigby: So, fair chance we're seeing something similar in the ac100 tree. Dec 16 23:19:20 jcrigby: Also, hey! Dec 16 23:19:40 jcrigby: Any plans to get linaro-lt-mx5 updated to support armhf before Christmas? Dec 16 23:20:34 I knew I should have remaind quiet Dec 16 23:21:00 * infinity laughs. Dec 16 23:21:02 when I think about that I think I should also move to newer kernel and that is more work Dec 16 23:21:13 jcrigby: I don't care too terribly much, I'm off until Jan 3. Dec 16 23:21:32 jcrigby: But the earlier we have everything working on armhf, the better. Dec 16 23:21:57 so is same kernel as you have now just with armhf an ok first step? Dec 16 23:22:06 if so easy Dec 16 23:22:35 jcrigby: Yup. Dec 16 23:22:44 ok then I am commited Dec 16 23:23:32 jcrigby: A second step would be to talk to markos about having an mx5 that we know works on mx53-loco and both efika platforms. No pressure. ;) Dec 16 23:23:40 (Cause I still intend to do hybrid boot media) Dec 16 23:24:21 swell Dec 16 23:29:01 janimo, so does cross building kernel work for you now, including tools? Dec 16 23:29:26 jcrigby, no, I did not get around to trying tools yet since other issues popped up Dec 16 23:29:31 like the kernel not booting at all Dec 16 23:29:55 jcrigby, I learned how to cope with ABI bumps though. more or less Dec 16 23:29:57 ok, I understand that would be a bigger issue Dec 16 23:32:15 oh, good. That hurt my brain for a long time. **** ENDING LOGGING AT Sat Dec 17 02:59:57 2011