**** BEGIN LOGGING AT Sun Jul 24 23:59:56 2005 Jul 25 03:34:37 DaKa2: ping Jul 25 04:21:57 rwhitby: pong Jul 25 04:25:15 I get some errors on ipkg install openslug-native Jul 25 04:25:45 dammit... Jul 25 04:26:05 http://pastebin.ca/18441 Jul 25 04:26:20 I haven't debugged it at all ... Jul 25 04:27:02 ahh, you need to: ipkg install busybox && ipkg install cpio && ipkg install sysvinit Jul 25 04:27:02 this was a clean openslug, then ipkg remove dropbear, ipkg install openssh, ipkg install rsync, modify cross-feed to point to my local latest openslug feed, ipkg install openslug-native Jul 25 04:27:43 Hmm. Our solution shouldn't require an order of installation or required installations. Jul 25 04:28:18 yes.. [g2] got this yesterday, I really dont know what to do Jul 25 04:28:29 I particularly don't want to install busybox Jul 25 04:28:41 ahh, then you dont need that Jul 25 04:28:51 if you dont have it already Jul 25 04:29:03 is it because the image this is coming from was pre- the changes? Jul 25 04:29:08 yes Jul 25 04:29:29 ok, so reflash with a latest image, then re-turnup -i , then try again .... Jul 25 04:29:34 ipkg upgrade should work Jul 25 04:29:40 but ipkg runs all the Configures at the end Jul 25 04:30:32 with the latest image everything works great Jul 25 04:30:45 ok, I'll reflash first. Jul 25 04:31:06 What you can do is make openslug-native depend on the latest image ipk. Jul 25 04:33:46 hm, you can? openslug-image is not in the list_installed on my slug, how to detect it? Jul 25 04:34:05 and.. we havn't bumped that PR for some time.. Jul 25 04:36:20 thinking ... Jul 25 04:58:26 hello all Jul 25 04:58:53 where should I make openslug feature requests like "could you enable ipv6 as a kernel modules?" ? Jul 25 05:30:22 koen: here :-) Jul 25 05:31:26 jbowler-away is the best person to talk to about that, but if you want to enable it in the defconfig and test that it doesn't break anything, then feel free to check it in Jul 25 05:32:05 it will only generate extra ipkgs Jul 25 05:32:47 yeah, should be a very benign change. The question is what else has to be modified to support it (i.e. busybox, etc) Jul 25 05:36:44 true Jul 25 05:36:54 I'll look into it after I finish some matlab work Jul 25 05:37:22 there was someone else here who was keen on ipv6 too, but I forget who it was. Jul 25 05:38:44 that was kolla_ Jul 25 05:40:13 It will also mean I have to dig up the miplv6 kernelpatch and see if that works Jul 25 05:49:28 mickey|meeting, Thanks for the udev fix. it worked fine. Jul 25 05:51:14 ipv6, yes - me Jul 25 05:51:56 I had plannet to write something clever about it, but since I havent done much else but enable the kernel module and giving the machines addresses.. Jul 25 05:52:20 busybox is obviously compiled with some limited ipv6 support Jul 25 05:52:40 I want to run tinc or miplv6 on it Jul 25 05:53:54 mickey|meeting: is the udev fix in OE yet Jul 25 05:54:02 ? Jul 25 05:54:17 * koen just killed his hx4700 with udev Jul 25 06:04:11 what is tinc? Jul 25 06:04:28 (or miplv6 for that matter) Jul 25 06:04:58 miplv6: http://www.mipl.mediapoli.com/ Jul 25 06:05:05 ah, mobile ip Jul 25 06:05:13 sorry, I should have recognized that Jul 25 06:05:27 tinc does something like that: http://www.tinc-vpn.org/ Jul 25 06:06:15 yes, sounds like nice things to use the slug for Jul 25 06:10:53 DaKa2: openslug-native fully installed. I added gzip to openslug-packages cause it was an unsatisfied recommendation for perl. Jul 25 06:12:48 also added monotone to openslug-native :-) Jul 25 06:14:48 koen, should be. My udev-063 works fine. Jul 25 06:14:58 ok, thanks Jul 25 06:15:04 But before it would give me errors on boot. Jul 25 06:15:59 the only problem right now with udev and hotplug-ng is that it doesn't pickup cold-boot devices. Jul 25 06:16:51 so in the case of my robot for example, the USB cam is always plugged in so on bootup it doesn't detect it unless I either force the modprobe or unplug and replug. Jul 25 06:17:02 not a big deal really though Jul 25 06:23:33 rwhitby: ah, great, cpan needs gzip, was going to do that but forgot about it Jul 25 06:25:04 just starting from fresh again to double check that there were no other warnings Jul 25 06:25:22 rwhitby: hm, you got perl working? Jul 25 06:25:31 rwhitby: gzip might need to be updated with u-a Jul 25 06:25:56 <[g2]> rwhitby, apex ??? Jul 25 06:26:05 yes Jul 25 06:26:19 <[g2]> are you running APEX ? Jul 25 06:26:37 hm, now monoton complains about "monotone: misuse: database schemas do not match:" Jul 25 06:26:42 [g2]: no, but it's in there in preparation for DebianSlug Jul 25 06:26:46 *shrug* Jul 25 06:26:59 kolla_: you need to update to 0.21 and follow the migration instructions "db migrate" Jul 25 06:27:12 <[g2]> cool Jul 25 06:27:15 update what to 0.21? Jul 25 06:27:16 <[g2]> I'll check the config Jul 25 06:27:19 monotone Jul 25 06:27:35 [g2]: it's direct from bewoolie's nslu2_config Jul 25 06:27:57 which I assume is from you Jul 25 06:28:18 <[g2]> yeah from us Jul 25 06:28:33 <[g2]> there's also an openslug_config Jul 25 06:28:44 <[g2]> which iirc it the one to use Jul 25 06:29:02 Anyone got a public dns-server I can use? Jul 25 06:29:10 <[g2]> i think the only difference is the kernel cmd line params Jul 25 06:29:23 repvik: 130.89.1.2 Jul 25 06:29:24 repvik: zoneedit Jul 25 06:29:48 it runs nslu2-linux and my other three domains for the past four years, for free. Jul 25 06:29:53 rwhitby: silly question.. how is one supposed to update monotone.. new Makefile form www.nslu2-linux.org? Jul 25 06:30:11 kolla_: you get monotone from your host distro Jul 25 06:30:20 right, ok Jul 25 06:31:00 it is 0.21 already, ok Jul 25 06:31:27 can anyone access ipkg.kynisk.com? Jul 25 06:31:59 I can, but it can't access anything outside its server room Jul 25 06:32:16 PING ipkg.kynisk.com (84.16.208.26): 56 data bytes Jul 25 06:32:20 and it just stays there Jul 25 06:32:33 (from my freshly flashed slug :)) Jul 25 06:32:38 strange Jul 25 06:33:09 because I can't reach destinations outside norway from that box Jul 25 06:33:33 congrats on the slug ;) Jul 25 06:34:08 rwhitby: "db migrate" where? Jul 25 06:34:50 monotone --db=yourdb db migrate Jul 25 06:35:33 hm, must look at this later.. Jul 25 06:36:06 it adds a few indices to speed things up a bit Jul 25 06:36:19 <[g2]> rwhitby, have you talked to Jacmet about getting DebianSlug in the repo at all ? Jul 25 06:36:25 yep Jul 25 06:36:36 and we have a wiki/DebianSlug/HomePage and wikigroup Jul 25 06:36:46 <[g2]> excellent Jul 25 06:37:22 and another guy (who's name escapes me at the moment) is working on recompiling the initrd so that it doesn't need a serial port. Jul 25 06:37:25 <[g2]> is he already working with beewoolie on a le_config ? Jul 25 06:37:59 not that I know of. But I have updated slugimage to do the byteswapping as the image is built, so it will work with stock redboot. Jul 25 06:38:03 rwhitby: in theory openslug should be debian compatible by installing apt-get, but I guess debian doesn't do armeb? Jul 25 06:38:18 koen: correct Jul 25 06:38:25 so, I take it that our ..uhm.. monotone repository has moved away from berlios? Jul 25 06:38:42 monotone never was at berlios Jul 25 06:39:01 we have moved the bitbake rev we use into our monotone repo Jul 25 06:39:17 s/moved/copied/ Jul 25 06:39:19 cause we couldn't get to bitbake for three days, we've added it to our monotone repo as org.nslu2-linux.bitbake, and will just update it there Jul 25 06:39:45 we're going to remove all build dependencies on berlios, cause they can't be bothered working on weekends, whereas we do. Jul 25 06:40:04 right Jul 25 06:40:15 jbowler is going to replace oe-symlinks with BBFILES and a freeze script to update it. Jul 25 06:40:31 * repvik looks forward to that Jul 25 06:41:19 [g2]: we will probably have to move to apex for DebianSlug due to redboot not being able to change the kernel command line Jul 25 06:41:40 <[g2]> right makes total sense Jul 25 06:41:41 otherwise we would have to build two debianslug images, one for install and one for running after that. Jul 25 06:41:59 and the second one would have a hard-coded rootfs. Jul 25 06:42:23 but we need to be able to get to apex console with a stock slug (no serial) - is that possible? Jul 25 06:42:32 hm, I must be dense do I did a "monotone --db=nslu2-linux.db db migrate" Jul 25 06:42:42 in the monotone directory Jul 25 06:42:50 kolla_: that should do it Jul 25 06:42:57 <[g2]> rwhitby, only possible when someone writes PCI/USB code for it Jul 25 06:43:08 <[g2]> PCI/USB/Serial Jul 25 06:43:18 rwhitby: guess I should also update my Makefile or something? Jul 25 06:43:39 kolla_: if you're using the master makefile, then make update ; make setup-bitbake Jul 25 06:43:45 ok Jul 25 06:44:10 oops, make that "make update-master ; make setup-monotone ; make setup-bitbake" Jul 25 06:44:44 hm Jul 25 06:44:47 monotone: misuse: no branch pattern given and no default pattern set Jul 25 06:44:48 update-master gets the new makefile, setup-monotone fixes your default include pattern to include org.nslu2-linux.bitbake, and setup-bitbake then gets bitbake from monotone. Jul 25 06:44:52 bails out Jul 25 06:45:01 ok, do setup-monotone first Jul 25 06:45:11 <[g2]> rwhitby, isn't LE support supposed to be in CSR 1.5 ? Jul 25 06:45:12 hmm. that won't work either Jul 25 06:45:21 yes, that at least does something :) Jul 25 06:45:23 wget a new Makefile, then do setup-monotone Jul 25 06:45:41 [g2]: yep, someone just needs to integrate it Jul 25 06:52:17 DaKa2: is anyone close to a cross-compiled perl which works? Jul 25 06:53:11 rwhitby: iirc it needed to be linked with libgcc to work Jul 25 06:54:50 DaKa2: "package ncurses suggests installing ncurses-terminfo" - should we add that to openslug-native ? Jul 25 06:55:47 It's not strictly necessary Jul 25 06:56:53 <[g2]> rwhitby, NAiL I built perl last night with openslug-native Jul 25 06:57:18 everything works ? Jul 25 06:57:31 <[g2]> pretty much Jul 25 06:57:33 rwhitby: is openslug-native the meta-package? Jul 25 06:57:40 <[g2]> yes Jul 25 06:57:52 <[g2]> heh.... koen :) Jul 25 06:58:22 * koen adds that to the 'steal from nslu2' list Jul 25 06:58:53 <[g2]> koen, consider it payback for all the support to get to here Jul 25 06:59:02 just don't take our fun-luvin' attitude :-) Jul 25 06:59:18 :) Jul 25 06:59:26 <[g2]> I thought that was ass-kick'n :) Jul 25 06:59:50 DaKa2: just completed a fresh flawless openslug-native install. Good job. Jul 25 06:59:57 <[g2]> koen, that was as a normal user also Jul 25 07:01:27 <[g2]> rwhitby was that just a ipkg update && ipkg insall openslug-native or did you need to install busybox and upgrade first ? Jul 25 07:04:38 built latest image from monotone, reflash -i image, turnup to a new disk, ipkg update from local feed, ipkg install openslug-native Jul 25 07:05:00 oh, brilliant Jul 25 07:05:01 <[g2]> rwhitby, nod... I'm reading the log now :) Jul 25 07:05:49 <[g2]> I built file natively, I'm not sure whether the static perl needs it, iirc the dynamic perl does Jul 25 07:06:01 <[g2]> and you'll need procps for the magic test to pass Jul 25 07:06:22 <[g2]> but otherwise you can just build perl Jul 25 07:09:26 <[g2]> koen, beewoolie wanted ipv6 Jul 25 07:09:46 so, openslug is now self-maintained on slug? :) Jul 25 07:10:34 <[g2]> koen, not yet but it probably could be Jul 25 07:10:56 <[g2]> s/koen/kolla/ Jul 25 07:11:22 Is there a problem with procps on openslug? It compiles fine on my end, but it's not in openslug-packages.bb Jul 25 07:11:30 (I can't test it right now unfortunately) Jul 25 07:14:14 I happened to compile procps when I was going to compile psmisc :P Jul 25 07:15:18 ~praise dyoung once again for fixing the shutdown issue Jul 25 07:16:54 hostname: too many arguments Jul 25 07:16:54 Try `hostname --help' for more information. Jul 25 07:17:04 looks like our hostname doesn't match the initscript Jul 25 07:19:52 <[g2]> nod on the hostname Jul 25 07:20:53 <[g2]> DaKa2, setup the update-alternative to use the busybox shutdown not the util-linux ones Jul 25 07:21:23 [g2]: where's your keys? ;-) Jul 25 07:22:24 <[g2]> I'll send them in soon Jul 25 07:22:45 <[g2]> DaKa2, did the proper thing as the util-linux shutdown doesn't work Jul 25 07:22:53 <[g2]> I wasn't complaining Jul 25 07:23:09 oh, the comma threw me Jul 25 07:23:41 as if you were addressing DaKa2 instead of referring to him :-) Jul 25 07:23:58 <[g2]> xchat autocomplete Jul 25 07:24:13 <[g2]> I hate that... there's probably and option to fix it Jul 25 07:25:00 <[g2]> I love the autocomplete 99% of the time, I hate the "," it puts in there Jul 25 07:27:14 NAiL: did you mt push without using "make push-openembedded" ? Jul 25 07:27:53 I never do just a push Jul 25 07:28:10 I pull, commit, sync Jul 25 07:28:15 Hmm - org.openembedded.nslu2-linux on vanille has two heads. Jul 25 07:28:29 do you do an update between the pull and commit ? Jul 25 07:28:37 yes Jul 25 07:28:56 I did make publish-openembedded after the sync Jul 25 07:28:56 and do you check for multiple heads like the Makefile does? Jul 25 07:29:16 As often as I can. I do forget it once in a while... Jul 25 07:29:38 I am pretty sure I did today though Jul 25 07:30:03 Hmm. we need to work out how those two heads got there, and update the Makefile so that using makefile commands *cannot* create two heads. Jul 25 07:30:09 actually, I don't check for heads, I just merge and ignore the error I usually get Jul 25 07:30:33 what error? Jul 25 07:30:39 misuse thingy Jul 25 07:30:44 there's only one head Jul 25 07:30:56 there is now, cause I just merged them. Jul 25 07:31:07 rwhitby: you can't garantee to only have one head Jul 25 07:31:18 yeah, must have been the race condition Jul 25 07:31:28 the makefile approach catches 99% of them, but not all Jul 25 07:31:30 NAiL: sorry about the inquisition :-) Jul 25 07:31:37 rwhitby: no probs Jul 25 07:32:39 If I do something incorrectly, I'd rather know than continue doing the same mistake and causing trouble ;-) Jul 25 07:32:50 hm, make setup-monotone seems to take forever.. Jul 25 07:33:00 how many revs are there to be written? Jul 25 07:33:09 kolla_: yes. Actually, it takes forever-4 hours Jul 25 07:33:11 <[g2]> 550 ? Jul 25 07:33:19 sheesh Jul 25 07:33:25 I cant stay here for 4 hours :) Jul 25 07:33:44 kolla_: heh, not 4 hours. Eternity minus 4 hours ;-) Jul 25 07:33:53 download http://ewi546.ewi.utwente.nl/OE/OE.db.bz2 and add your privkey Jul 25 07:34:04 I broke it and now run itside screen Jul 25 07:34:12 that's a 3 hour old bzipped monotonedb without privkeys Jul 25 07:34:32 kolla_: dyoung-zzzz is working on a cron job for the same thing for nslu2-linux. then we will mod the makefile to use that as the starting point. Jul 25 07:35:06 why does it take so long anyways? Jul 25 07:35:11 looks like the slug takes a couple of minutes for each rev. Jul 25 07:35:17 lots of crypto Jul 25 07:35:24 what for? Jul 25 07:35:35 all those keys and hashes Jul 25 07:36:52 *shrug* Jul 25 07:36:57 apparently monotone is designed to be tamper-proof in the face of an insecure transport medium and hostile intermediary netsyncs Jul 25 07:37:13 correct Jul 25 07:37:14 rwhitby: I dont think ncurses-terminfo is needed.. Jul 25 07:37:24 but only 0.21 fixes a man in the middle scenario Jul 25 07:37:32 or 0.20 Jul 25 07:37:49 oh well.. I'll let it run along then, and do something useful instead in the meantime Jul 25 07:37:56 like.. dinner Jul 25 07:38:03 "make setup" is now checking out the openembedded dir on the slug, ready for a native build ... Jul 25 07:38:40 fantastic! Jul 25 07:39:21 just run make setup, then Ctrl-C it while it's downloading the repo, then copy a pre-built one over, then run make setup again Jul 25 07:39:49 once dyoung-zzzz has the cron job done, then it will just download the initial repo anyway for cross and native builds. Jul 25 07:39:54 oh, and perl, we have to either fix the oe perl .bb, or add our own for native building Jul 25 07:40:05 rwhitby: have another financing round, buy 40 slugs and install icecream fur distributed builds ;) Jul 25 07:40:54 <[g2]> looks like procps is fixed in the repo Jul 25 07:41:10 <[g2]> I just built/install and ran the perl test and it passed Jul 25 07:41:28 s/fur/for/ Jul 25 07:42:02 [g2]: procps works cross? Jul 25 07:44:14 DaKa2: ok, now we need openslug-native.conf - do you have one of those? Jul 25 07:44:14 DaKa2: getting a __umodsi3 problem? Jul 25 07:45:05 <[g2]> NAiL, looks like it Jul 25 07:45:15 <[g2]> at least enought for the perl test to pass Jul 25 07:45:45 ok, adding it back into packages then Jul 25 07:46:35 DaKa2: "/bin/sh: source: not found" Jul 25 07:47:01 DaKa2: http://bugs.openembedded.org/show_bug.cgi?id=113 Jul 25 07:47:04 Hmm - I guess I need to be running bash :-) Jul 25 07:47:28 some meta package that removes busybox and replaces it with fully functional programs would nice :) Jul 25 07:47:53 DaKa2: and I remember adding libgcc to LD_PRELOAD fixes the missing symbol problem (if that is the perl problem you're facing) Jul 25 07:48:27 rwhitby: sed -i s/source/\./g * Jul 25 07:48:34 koen: ah, yes, I know, Ive fixed that in the .bb Im using on the slug Jul 25 07:48:38 rwhitby: bash Jul 25 07:48:43 ah Jul 25 07:48:45 yes Jul 25 07:50:30 rwhitby: we had problem with 'source' when adding the maemo distribution. We patches all the files to use '.', but one could teach busybox to understand 'source' Jul 25 07:51:42 ok, let's try the first commit from a slug then .... Jul 25 07:52:00 <[g2]> koen, how is the maemo distro looking ? August is just around the corner :) Jul 25 07:52:51 [g2]: we have a functional maemo-image in OE, but we need to update it once the updates hit the street Jul 25 07:53:18 * [g2] is looking forward to the 770 :) Jul 25 07:54:00 * koen too Jul 25 07:54:20 * mickeyl three Jul 25 07:55:26 * rwhitby doesn't think the 770 will have enough over the Treo650 for him to buy one with his own money ... Jul 25 07:55:53 it lacks a phone for example Jul 25 07:56:11 and no qwerty keyboard for irc Jul 25 07:56:23 correct Jul 25 07:56:33 * koen starts saving for a bluetooth keyboard Jul 25 07:56:59 koen: might as well just run linux on a tablet PC ... Jul 25 07:57:27 * rwhitby waits for CIA to notify the very first monotone commit and push from a slug ... Jul 25 07:58:23 rwhitby: I can't get a tablet for $99, but yeah, it's about the same Jul 25 07:58:47 koen: how many can you get at $99 ? Jul 25 07:58:54 one Jul 25 07:59:22 the dev discount program is personalized Jul 25 07:59:58 there it is! the slug is now self-hosting CM-wise for monotone. Jul 25 08:00:19 congrats Jul 25 08:00:21 pretty cool Jul 25 08:00:24 rwhitby: congrats Jul 25 08:00:40 thank jbowler-away - he added the package Jul 25 08:00:54 jbowler-away: congrats & thanks Jul 25 08:10:35 NOTE: Handling BitBake files: \ (0004/0658) [ 0 %] Jul 25 08:10:52 <[g2]> take about 20-25 minutes Jul 25 08:10:55 heh - do we need a psyco package for the slug ? Jul 25 08:11:26 I didn't look much at it, gave me a segfault when I installed it Jul 25 08:12:41 but psyco will give speedups at the expense of memory, no? Jul 25 08:13:30 which means more swapping Jul 25 08:14:02 my idea was to just do a quick test and see if it was worth it, but since it didn't work at first attempt I skipped that Jul 25 08:14:17 ok, we will need to work out how to turn off the warning for native builds so that we don't get people continually asking about it Jul 25 08:16:56 <[g2]> rwhitby, the python install isn't complete Jul 25 08:17:40 <[g2]> just FYI Jul 25 08:18:07 <[g2]> some packages like Zope didn't build properly with the fully python install Jul 25 08:18:21 * rwhitby starts a "while true ; do make update ; done" loop on the slug to keep it up to date overnight ... Jul 25 08:18:58 [g2]: Zope's not in openslug-packages, so it doesn't exist for me yet :-) Jul 25 08:19:21 night all Jul 25 08:19:29 <[g2]> nite Jul 25 08:19:32 nite Jul 25 08:19:41 nite Jul 25 08:19:54 'night rwhitby-asleep Jul 25 10:29:54 Irssi 0.8.10-rc5 (20050724 2106) - running on Linux armv5teb - http://irssi.org/ Jul 25 10:29:57 \o/ Jul 25 10:30:46 yay :) Jul 25 10:30:58 <[g2]> irssi from the slug ? Jul 25 10:31:01 I didn't get irssi running in my build Jul 25 10:31:03 yes :) Jul 25 10:31:10 congrats :) Jul 25 10:31:16 i've built it on the box Jul 25 10:31:17 Now we've got screen and irssi running :D Jul 25 10:31:32 ah, native Jul 25 10:31:36 That doesn't count :P Jul 25 10:31:42 pah :P Jul 25 10:32:10 well it hardly depends on the glibc-gconv and glibc-charmap stuff :) Jul 25 10:34:02 <[g2]> senneth, did you use the openslug-native package ? Jul 25 10:34:12 yes Jul 25 10:34:39 I guess irssi will show up in the native feed eventually? :) Jul 25 10:34:59 :) Jul 25 10:35:13 you need pkg-config for it Jul 25 10:35:25 i first compiled pkg-config Jul 25 10:36:02 [g2]: is openslug-native ready to be installed on banana? Or is it already there? Jul 25 10:37:52 and irssi crashed all the time, it stopped after i did ipgk list|grep glibc-gconv|awk '{print $}'|xargx ipkg install and ipgk list|grep glibc-charmap|awk '{print $}'|xargx ipkg install Jul 25 10:37:55 <[g2]> NAiL, dunno what the plan is but openslug-native is shaping up well Jul 25 10:38:11 s/xargx/xargs/ Jul 25 10:38:53 brb, installing screen :) Jul 25 10:41:05 senneth: how much mem does it use? Jul 25 10:42:08 have to check that, i think it will use as much as possible :/ Jul 25 10:42:38 PID USER STATUS RSS PPID %CPU %MEM COMMAND Jul 25 10:42:40 823 vb S 3688 818 0.0 12.0 irssi Jul 25 10:46:53 what's the recommended way to 'switch' disks? I turnup'ed a flashdrive and now I want to use a real hda as root, which currently is on sdb but is going to be sda Jul 25 10:47:59 <[g2]> koen, you've got a couple choices Jul 25 10:48:20 <[g2]> are you planning on using the flash or are you migrating to hd Jul 25 10:48:44 I want to use the hd instead of the usbkey Jul 25 10:49:00 <[g2]> is there much stuff on the flash stick ? Jul 25 10:49:29 not really, couple of kernel modules Jul 25 10:49:41 nothing I couldn't reinstall in 3 ipkg commands Jul 25 10:49:56 <[g2]> then you could just turnup again to the disk Jul 25 10:50:03 turnup init, switch disk and turnup disk? Jul 25 10:50:34 <[g2]> you probably don't need to init just Jul 25 10:50:37 <[g2]> shutdown Jul 25 10:50:53 <[g2]> reboot without any drives attached Jul 25 10:51:14 <[g2]> plug in the hd Jul 25 10:51:18 <[g2]> format it Jul 25 10:51:36 <[g2]> turnup disk -i /dev/sda1 -t ext3 Jul 25 10:52:23 mother of all sweetness Jul 25 10:52:26 that's easy Jul 25 10:52:40 <[g2]> do you know about reflash ? Jul 25 10:53:08 with upslug or 'live' ? Jul 25 10:53:13 live Jul 25 10:53:21 <[g2]> from the cmd line Jul 25 10:53:36 <-- impressed with the work being done Jul 25 10:53:44 that shouldn't be too hard Jul 25 10:53:44 <[g2]> you can copy an image over and "reflash -k kernel.img" Jul 25 10:53:58 thought so Jul 25 10:54:07 <[g2]> or "reflash -i image" to replace the kernel and jffs2 Jul 25 10:54:25 * koen ponders about using xfs instead of ext3 Jul 25 10:54:35 <[g2]> jbowler runs reiserfs Jul 25 10:55:04 <[g2]> you'll need to be a little careful as the jffs2 is getting pretty full Jul 25 10:55:16 reiserfs pissed me off too many times Jul 25 10:55:20 well xfs is memory expensive Jul 25 10:55:20 <[g2]> you may want to remove reiserfs from the .conf Jul 25 10:55:30 Hans Reiser: "If you pay me, I'll break compat" Jul 25 10:55:32 xfs has pissed me of a bunch of times Jul 25 10:55:41 I've lost bucketloads of data due to xfs screwing up Jul 25 10:56:55 <[g2]> turnup works with nfs too Jul 25 10:57:07 the somewhat harmless but very annoying thing was when I was trying to configure XFree. Edit config-file, try, crash. When I booted I consistently found my config-file filled with ^@ Jul 25 10:57:21 * koen mumbles something about xfs+nfs on 2.6 bugs Jul 25 10:57:38 xfs and nfs doesn't play nice on 2.6, does it? Jul 25 10:57:39 yeah that happened with my .irssi/config aswell Jul 25 10:57:47 i switched to ext3 after it Jul 25 10:58:05 and my .irssi/default.theme was b0rked aswell :( Jul 25 10:58:11 I didn't switch, and it ended up picking file after file to destry when my box crashed Jul 25 10:58:34 Switching to anything else fixed that ;-) Jul 25 10:58:55 I filed a bug too, but never got any reply Jul 25 10:59:52 mem usage now? Jul 25 11:00:06 mmm, irssi is using ~3mb screen is using ~1mb Jul 25 11:00:36 well, more likely ~4mb Jul 25 11:00:44 btw, openslug is too well documented and too easy to use for a hobby FOSS project Jul 25 11:01:01 <[g2]> heh :) Jul 25 11:01:05 koen: yes. We should remove all the docs ;-) Jul 25 11:01:22 <[g2]> proof that FOSS is going mainstream :) Jul 25 11:01:34 <[g2]> with or without the devs Jul 25 11:01:46 and autoformat hardrives to UFS Jul 25 11:01:48 I think I could get my mom to install openslug without too much work Jul 25 11:02:13 <[g2]> koen, I'm happy you like it Jul 25 11:02:22 go for a single click installation ;) Jul 25 11:02:24 could you rename turnup to turnip? Jul 25 11:02:32 would save me some typos Jul 25 11:02:39 heh Jul 25 11:02:46 <[g2]> that was dyoung-zzzz joke Jul 25 11:02:49 and it would give you a reference to Blackadder Jul 25 11:04:11 * koen thinks off a way to explain the "big noisy disk" attached to the "cute silent internet thingy" to his girlfriend Jul 25 11:04:44 <[g2]> get a Maxtor -- set the spin down to 5 min :) Jul 25 11:04:52 exchange the disk by a vibrator and you will win ;-) Jul 25 11:05:41 oh the time is wron on the slug :) Jul 25 11:05:42 get a seagate Jul 25 11:06:01 <[g2]> TZ and ntp are your friend Jul 25 11:06:08 then adjust the noise/speed tradeoff thingy Jul 25 11:06:18 makes my hdd virtuall silent Jul 25 11:06:24 s/ll/lly/ Jul 25 11:06:55 <[g2]> senneth, so are you going to add an OpenSlug irssi/screen how to ? :) Jul 25 11:07:07 well i've an external case but it's made of plastic and it doesn't have any fans, and i think the hd will die from heat in some time :) Jul 25 11:07:22 plastic? Jul 25 11:07:25 that sound nice Jul 25 11:07:46 heh Jul 25 11:07:59 the 'case' of the usb hdd got lost months ago Jul 25 11:08:07 haha Jul 25 11:08:07 PSU + PCB + disk Jul 25 11:08:23 I'm considering the same actually Jul 25 11:08:36 i'll just put the electronics out of the case into a metalbox with a slow 120cm fan =) Jul 25 11:08:38 Or, rather build my own Jul 25 11:10:19 well my coworker builds anykind of cases, so i've to spent some beer to get a nice box :) Jul 25 11:11:32 jbowler-away: what was the setting in local.conf to do a uclibc build again? Jul 25 11:14:45 NAiL: TARGET_OS_local = "linux-uclibc" Jul 25 11:15:06 I'm going to add a .conf for this - because the result definately isn't "openslug" Jul 25 11:17:35 <[g2]> isn't "openslug" anymore Jul 25 11:18:19 ? it never was - because it was never possible to install 'openslug' packages on it Jul 25 11:18:44 uSlug? Jul 25 11:18:52 yay Jul 25 11:18:59 * koen too lazy to search for the 'mu' char Jul 25 11:19:07 only three distros to keep track of ;) Jul 25 11:19:29 only if you want... Jul 25 11:19:43 I've been wondering how to handle that actually Jul 25 11:19:52 Because the feeds are different Jul 25 11:19:57 but not much else, hopefully Jul 25 11:20:41 Yes, but I don't image a binary release, so anyone using it will have to build it... Jul 25 11:21:21 <[g2]> real-men and women build from source :) Jul 25 11:21:44 no, they enter the bits directly from JTAG Jul 25 11:24:10 reminds me of the place I worked five years ago... where one of they guys fixed the filesystem from the bios by twiddling bytes Jul 25 11:24:20 that wasn't exactly x86 ;) Jul 25 11:24:47 x86 sucks Jul 25 11:24:52 One of the kerneli guys, iirc Jul 25 11:25:18 I was surprised how well designed sun and sgi hardware was when first encountering it Jul 25 11:25:41 yup Jul 25 11:25:47 <[g2]> yeah I hear sun and sgi are really rock'n these days :) Jul 25 11:25:49 a great deal more thought were put into that hw Jul 25 11:26:17 <[g2]> chapter-12 ? :) Jul 25 11:26:32 It's a shame Jul 25 11:26:42 <[g2]> like HP Jul 25 11:27:02 hmm Jul 25 11:27:14 what was that range of cpus that died four-five years ago? Jul 25 11:27:29 68000 Jul 25 11:27:35 <[g2]> heh Jul 25 11:27:39 nah, still using 68k ;-) Jul 25 11:27:49 or SPARC - are SUN still doing SPARC? Jul 25 11:28:21 or 32016, going back a bit longer though. Jul 25 11:28:40 alpha :) Jul 25 11:28:42 or i860 - which was still-born Jul 25 11:28:52 <[g2]> Alpha was a DEC thing Jul 25 11:28:55 yeah Jul 25 11:29:07 lots of dead CPU families around Jul 25 11:29:17 sun is doing opteron now Jul 25 11:29:19 <[g2]> You're thinking the HP processor before Itanic Jul 25 11:29:31 * koen looks at his sparcs, sgi's and hp xterminals Jul 25 11:29:58 * koen is annoyed that the slug runs circles around his sparc5 Jul 25 11:30:13 <[g2]> koen, I'm getting in the hw biz Jul 25 11:30:39 PA-RISC Jul 25 11:30:41 that was fun Jul 25 11:30:50 * NAiL loathes HP/UX Jul 25 11:31:00 <[g2]> that's it Jul 25 11:31:09 * koen looks at the his apollo 715/75 Jul 25 11:31:28 <[g2]> Soul of a new machine -- again Jul 25 11:31:29 surprising how much hw I bought for E2,- :) Jul 25 11:31:46 * DaKa2 thinks of his 816/E36 (i think) Jul 25 11:32:10 <[g2]> Hey I spent $3K+ for my first 16Mhz 286 Jul 25 11:32:47 That's probably how much I could pay to get that old cray that is (virtually) next door Jul 25 11:33:15 <[g2]> If I only went long MSFT --- I'd be on some beach right now .... Jul 25 11:33:42 I'd like to have a cray to play with Jul 25 11:33:51 <[g2]> Ahhh... it's much better to be noble and battling *evil* Jul 25 11:34:02 only problem is that it isn't exactly embedded :-P Jul 25 11:35:09 hmm... I wonder if I can get the slug to be a server for my X88-terminal Jul 25 11:41:20 I read something about adapting packages to carry webtool modules? Jul 25 11:41:31 hmm Jul 25 11:41:38 kinda sounds like a neat ide Jul 25 11:41:40 a Jul 25 11:41:51 couldn't webtools parse ipkg status and activate internal modules? Jul 25 11:41:51 if webtool is installed, that is Jul 25 11:42:19 adapting hundreds of packages instead of one seems a bit backwards Jul 25 11:42:48 How much space do these modules take anyway? Jul 25 11:43:15 a few kb at most I guess Jul 25 11:44:47 I think webtools would gain more traction if you aren't forced to reinstall your system to get some 3rd party modules Jul 25 11:45:18 as the proposal would imply, which is why I asked Jul 25 11:45:42 <[g2]> koen, I'm thinking of putting a web front-end on ipkg Jul 25 11:46:13 <[g2]> is there a url for webtool ? Jul 25 11:46:28 heh Jul 25 11:46:28 webtools.wl500g.info/ Jul 25 11:46:39 I got that url from the nslu2 mailing list :) Jul 25 11:47:15 [g2]: go dog! Jul 25 11:47:19 <[g2]> I looked at that a couple months ago Jul 25 11:47:26 <[g2]> hey beewoolie-afk ! Jul 25 11:47:55 <[g2]> I wasn't really jazzed about it Jul 25 11:48:09 <[g2]> it straight forward but I think really too complex Jul 25 11:48:29 http://article.gmane.org/gmane.comp.misc.nslu2.linux/7641 Jul 25 11:49:04 <[g2]> beewoolie-afk, hey dude! .... I wanted to talk to you about APEX and LE, I think there's gonna be DebianSlug support Jul 25 11:49:32 * koen is still debating wether to install apex Jul 25 11:49:43 <[g2]> APEX ROCKS! Jul 25 11:49:52 I've been reading about that. do you mean there is going to be BE ARM in debian? Jul 25 11:50:11 <[g2]> no it runs LE need an ethernet adapter Jul 25 11:50:12 BTW, I'm making progress on the network stack in APEX. Jul 25 11:50:30 <[g2]> I saw there was a .19 rel Jul 25 11:50:34 Ah. What is special about Debianslug then? Jul 25 11:50:56 Different from other ARM stuff? Jul 25 11:51:08 <[g2]> from my mind it's just an APEX tweek and maybe a little toolchain stuff to build a flashable image Jul 25 11:51:40 <[g2]> right now everything seems hacked to work (which is very cool) but it's really overkill Jul 25 11:52:07 <[g2]> there's a byte-swapping program and all kinds of stuff Jul 25 11:52:55 <[g2]> I'd think it'd be a few instructions in the hw init and boom we'd be in LE mode Jul 25 11:53:12 Hmm. I don't think I understand the issues. Shouldn't we be able to boot Debian as is? The CPU is LE by default. All we should have to do is change the boot loader and we can go crazy. Jul 25 11:54:34 <[g2]> Hmm... it really comes up LE ? Jul 25 11:55:23 <[g2]> The would make it even easier Jul 25 11:55:40 <[g2]> just build a straight LE toolchain Jul 25 11:55:45 Yep. APEX changes to BE first thing, but it doesn't really need to. Jul 25 11:56:06 The BE/LE difference isn't that important to most software. Jul 25 11:56:16 For example, CPU registers are agnostic. Jul 25 11:56:44 <[g2]> but what's stored in memory matter :) Jul 25 11:57:19 <[g2]> So if you can could you create a le_config in mach-ixp4xx ? Jul 25 11:57:24 Not exactly. Jul 25 11:57:38 I mean, yeah, it's easy to build APEX in LE mode. Jul 25 11:58:00 <[g2]> well we should talk to Jactmet in #nslu2-linux about it Jul 25 11:58:03 The not exactly is that the memory layout isn't different. The bytes are stored the same way in physical ram either way. Jul 25 11:58:20 It's only a problem when you try to read bytes. Jul 25 12:01:21 perl -MCPAN -e install Bundle::CPAN is killing everything :) Jul 25 12:02:08 d'oh Jul 25 12:02:09 <[g2]> senneth, did you build your own perl or install ? Jul 25 12:02:15 forot to mount the swap :) Jul 25 12:02:24 [g2]: just the installed version Jul 25 12:03:02 <[g2]> senneth, so can you do that howto on the wiki ? Jul 25 12:03:24 later, just trying to get some scripts to work with irssi Jul 25 12:03:44 <[g2]> Great thx! Jul 25 12:05:56 ahh, it didn't liked the swap partition i made with gentoo Jul 25 12:06:50 <[g2]> senneth, it didn't like the swap partition I created on the x86 either :) Jul 25 12:07:01 ohh, ic :) Jul 25 12:07:05 swap isn't BE safe Jul 25 12:07:08 ? Jul 25 12:07:12 prolly Jul 25 12:07:15 <[g2]> BE is safe Jul 25 12:07:23 <[g2]> but the x86 created a LE Jul 25 12:07:35 <[g2]> iirc bad magic Jul 25 12:08:05 <[g2]> there's a funny note in there HowTo NFSSwap I wrote up Jul 25 12:08:14 <[g2]> in the Jul 25 12:08:17 Unable to handle swap header version 16777216 Jul 25 12:08:25 it doesn't like the header Jul 25 12:08:31 <[g2]> mkswap Jul 25 12:08:39 i did already :) Jul 25 12:08:47 i'm just pasting from dmesg Jul 25 12:08:55 <[g2]> did you zero out the partition ? Jul 25 12:09:07 not really, just mkswap Jul 25 12:09:53 <[g2]> you can probably just zero out the first 10K or something Jul 25 12:10:59 perl consumes all the memory, i think running irssi with perl script will make the box swapping all the time Jul 25 12:11:18 can you run irssi without perl? :) Jul 25 12:11:39 sure Jul 25 12:11:51 probably a stupid question: is openslug-native in the feed yet? Jul 25 12:11:57 ./configure --without-perl :) Jul 25 12:12:06 afaik, perl is in the native feed Jul 25 12:12:14 duh Jul 25 12:12:27 koen: I'm rebuilding the feed now Jul 25 12:12:29 yes and ./configure finds it and it compiles it in Jul 25 12:12:47 NAiL: thanks Jul 25 12:12:55 <[g2]> senneth, my prototypes should arrive in a couple weeks Jul 25 12:13:04 <[g2]> they've got 64MB on board Jul 25 12:13:13 lucky you Jul 25 12:13:39 <[g2]> Production it's long after that and I'll be selling them Jul 25 12:14:03 i already made a turboslug out of mine Jul 25 12:14:09 <[g2]> Cool little unit, dual ethers, POE, CF slot Jul 25 12:14:11 and prepared it for a serial port Jul 25 12:14:18 <[g2]> serial, JTAG Jul 25 12:14:50 <[g2]> just FYI Jul 25 12:14:58 mmm, dual ethers Jul 25 12:15:06 yummie Jul 25 12:15:12 <[g2]> POE Jul 25 12:15:22 oh yes Jul 25 12:15:31 more power to you! Jul 25 12:15:32 poe? what does it stand for? Jul 25 12:15:38 power over ethernet Jul 25 12:15:42 <[g2]> nod Jul 25 12:15:42 oh ic Jul 25 12:16:03 the Gbit POE specs are funny Jul 25 12:16:11 as you can imagine Jul 25 12:17:50 [g2]: ARM boards or mips? Jul 25 12:18:28 <[g2]> koen, IXP422 like slug but has MD5/SHA-1 in the NPEs Jul 25 12:18:50 <[g2]> it runs at 266Mhz with 64MB out of the box Jul 25 12:18:51 [g2]: faster monotone! Jul 25 12:19:12 <[g2]> actually, something like that would be cool Jul 25 12:19:29 <[g2]> but it'll require some tweaking/building of the NPE libs Jul 25 12:26:38 is david.thg.se still the correct source for the native feeds? Jul 25 12:29:06 <[g2]> koen, that was just a test of native perl Jul 25 12:29:14 <[g2]> we're working on the native feed Jul 25 12:29:33 I guessed the wiki would be out of date :) Jul 25 12:29:45 since you don't update it every 10 minutes Jul 25 12:29:50 <[g2]> with the openslug-native package in the repo we are getting a log closer Jul 25 12:29:54 <[g2]> a lot Jul 25 12:30:14 <[g2]> rwhitby-asleep, actuall ran monotone on the slug today Jul 25 12:30:38 <[g2]> and commited a check in Jul 25 12:30:39 I was there to see the CIA messages roll by Jul 25 12:30:44 <[g2]> nod Jul 25 12:42:34 <[g2]> senneth, around ? Jul 25 12:42:39 yes Jul 25 12:42:55 <[g2]> I built irssi how do I run it ? I haven't installed it Jul 25 12:43:09 src/fe-text/irssi Jul 25 12:44:35 <[g2]> I guess I didn't build that with the default config :) Jul 25 12:44:55 <[g2]> it's building now Jul 25 12:45:36 have you installed pkg-config or did you let irssi build a glib-1.2? Jul 25 12:45:54 <[g2]> I think it built glib-1.2 Jul 25 12:46:33 <[g2]> yeah 1.2.10 to put a point on it Jul 25 12:47:19 <[g2]> ooops....make: *** No rule to make target `../fe-common/irc/libfe_common_irc.a', needed by `irssi'. Stop. Jul 25 12:48:00 and FYI i've build the svn head Jul 25 12:48:49 <[g2]> thx I'm attempting irssi-0.8.9 Jul 25 12:49:13 and please compile and install pkg-config first and glib-2 :) Jul 25 12:49:23 so it won't mock you with glib-1.2 Jul 25 12:49:32 and glib2-deb Jul 25 12:49:35 err -dev Jul 25 12:50:11 and the glibc-gconv and glibc-charmap packages aswell, or it will crash on you Jul 25 12:50:34 <[g2]> well I'll wait a while Jul 25 12:50:59 <[g2]> I thought it was worth a try to see if it'd configure and run Jul 25 12:51:05 i hope my memory serves well wehn i write the how-to Jul 25 12:51:31 <[g2]> that's ok... get getting a start then ppl can follow and correct Jul 25 12:51:41 <[g2]> or get stumped for a little while Jul 25 12:55:57 i wonder how to get a user on the wiki *rolleyes* Jul 25 12:56:24 <[g2]> where are you trying to update ? Jul 25 12:57:01 i think it would fit into Native Development Jul 25 12:57:40 MakeIrssiOnOpenSlug or something like that Jul 25 12:57:58 <[g2]> sounds good Jul 25 12:58:16 <[g2]> you shold be able to create it of the HowTo page Jul 25 12:58:29 well, it asks for a password :) Jul 25 12:59:55 and i think I've to toy a little with the wiki-sandbox to figure out the textformats Jul 25 13:00:26 just enter the name of the page you want to create Jul 25 13:00:46 Ok Jul 25 13:01:38 The easiest way is to take an existing page in the right place and change the last element of the URL to something appropriate. Jul 25 13:06:34 * slugirssi tries /sysinfo ;) Jul 25 13:06:43 Host 'cheiron', running Linux 2.6.12.2 -; Free: [Mem: 1/30 Mio] [Swap: 188/245 Mio]; Vpenis: 8 cm; Jul 25 13:07:35 it highly depends on a real version of df ... Jul 25 13:08:21 and scripts work if you have the appropriate perl modules installed Jul 25 13:17:56 hm.. /sysinfo? Jul 25 13:19:19 oh.. Jul 25 13:19:24 Host 'hem', running Linux 2.4.27-2-686-smp - Cpu0: Intel 2399 MHz Cpu1: Intel 2399 MHz Cpu2: Intel 2399 MHz Cpu3: Intel 2399 MHz; PCI: VGA,IDE interface: ServerWorks CSB6 RAID/IDE,USB,Ethernet,Ethernet; Up: 90d+2:41; Users: 9; Load: 0.00; Free: [Mem: 18/1009 Mio] [Swap: 2008/2048 Mio] [/: 3097/5629 Mio] [/home: 49656/63831 Mio]; Vpenis: 58.8 cm; Jul 25 13:19:55 hm.. a bit more output.. Jul 25 13:20:21 Host 'knulp', running Linux 2.6.12 - Cpu0: Celeron 468 MHz Cpu1: Celeron 468 MHz; Up: 5d+4:10; Users: 6; Load: 0.06; Free: [Mem: 7/374 Mio] [Swap: 256/259 Mio] [/: 2475/15002 Mio] [/var: 1255/4116 Mio] [/data: 15714/42934 Mio] [/boot: 6/23 Mio]; Vpenis: 29.3 cm; Jul 25 13:20:29 that's my normal gateway Jul 25 13:21:57 I just use 'hem' for my screened irssi Jul 25 13:22:29 other than that its a webserver for ~1600 virtual hosts Jul 25 14:30:12 ok, lets see, perl 5.8.7 crosscompiled.. ipkg list | grep ^perl | grep 5.8.7-r0 | cut -f1 -d" " | xargs -n 100 ipkg install Jul 25 14:30:44 just 858 packages... Jul 25 14:31:43 I remember jacques saying something like "holy crap, it generated 700+ packages" Jul 25 14:32:10 yep, my .bb Im building nativly geenrates 5 Jul 25 14:32:28 basically perl, perl-modules, libperl, and some other Jul 25 14:35:11 the perl packaging is something that really needs to be discussed sometime, 858 packages is just wrong Jul 25 14:36:44 it is granular Jul 25 14:36:51 but we could do some grouping Jul 25 14:36:51 hehe, yes :-) Jul 25 14:40:27 the ipkg state info for all the 858 packages takes 6,8Mb :) Jul 25 14:41:42 heh Jul 25 14:41:52 kinda defeats the purpose Jul 25 14:42:02 but not all info in /status is needed Jul 25 14:42:33 actually that was /usr/lib/ipkg/info/perl* Jul 25 14:42:45 DaKa2, which is why if you do a ipkg update without a HD attached it kills the flash Jul 25 14:43:32 ka6sox-office: it shouldn't, that should be put in /var/ipkg and when on flash that should be tmpfs Jul 25 14:44:12 then its been fixed...because before it would fill the flash and kill it. Jul 25 14:44:23 hm. maybe it does.. Jul 25 14:44:26 (I think that is on Unslung) Jul 25 14:44:42 ah, Im just using openslug Jul 25 14:44:50 k Jul 25 14:44:58 we did chance that in OE, but I'm not sure for which devices Jul 25 14:45:43 at 6.8mb of data its too much for most flash devices. Jul 25 14:46:34 the side effect of ultra granularity Jul 25 14:46:44 :) Jul 25 14:46:55 yeah, well, this is just when installed Jul 25 14:47:32 25Mb of perl, 6,8Mb of /usr/lib/ipkg/info/perl* Jul 25 14:48:15 ka6sox-office: you cannot run ipkg without unslinging first under unslung (now). Jul 25 14:48:57 w00t..its fixed! Jul 25 14:49:25 if you unsling and then the HD is disconnected you can still do a ipkg update. Jul 25 14:49:33 and it will fill the flash...iirc. Jul 25 14:49:47 probably. Jul 25 14:49:51 or is ipkg in the HD only and not in the flash? Jul 25 14:50:34 possibly. Jul 25 14:50:45 then you are safe... Jul 25 14:50:49 I'm not running unslung actively anymore, so I dunno. Jul 25 14:51:28 My last act of unslung stuff is integrating zd1211 into the build system; but i cant do that yet because theres something funny with glibc. Jul 25 14:52:11 IIRC the fix used was to make the ipkg on the flash file system non-functional, but I can't remember the details. Jul 25 14:54:00 I think we provide the symlink for "ipkg" in unsling now. Jul 25 14:54:01 jbowler-away, that would work. Jul 25 14:54:36 so if you try to run "ipkg" prior to unslinging, it tries to run the directory named ipkg in /etc , giving a error. Jul 25 14:55:16 jbowler-away: since you're around now, do you recall what you did to make that glibc / glibc-intermediate message go away around a week ago? Jul 25 14:59:53 I didn't do anything - it's still there so far as I know. It always has been there... Jul 25 15:00:32 We don't need an intermediate glibc, because glibc is not required to build the kernel Jul 25 15:00:46 (Or vice versa) Jul 25 15:01:21 glibc-intermediate is there, I believe, to fix a bootstrap problem with a kernel which requires a glibc and a glibc which requires kernel headers Jul 25 15:01:46 I see. Jul 25 15:02:04 Instead we build gcc-initial, which uses linux-libc-headers and is then used to build the kernel. Jul 25 15:02:25 I'll try a blowaway build later and see what the specific error is... Jul 25 15:02:55 In fact this is slightly broken - there are excessive dependencies in the kernel.bbclass - I fixed those for bootbox, but they are major core changes so I never checked them in. Jul 25 15:04:00 The message is harmless, because it choses the correct one, and a simple PREFERRED_PROVIDER would make it go away, though it would take some experiment to get the right value for PREFERRED_PROVIDER... Jul 25 15:04:37 Hmm, I see. Jul 25 15:04:50 Thanks! Jul 25 15:05:01 ... Back Later. Jul 25 15:30:34 night all Jul 25 15:30:41 nite Jul 25 15:32:01 "Failed 7 test scripts out of 223, 96.86% okay." not too bad for a first try of crosscompiled perl Jul 25 15:32:18 most of the problems are because of my way of testing anyway.. Jul 25 15:35:33 if anyone know a good supported way of running the perl testsuite on an installed system, please tell me.. Jul 25 15:44:09 when you see fontenot ask him. Jul 25 15:48:10 ok, thanks Jul 25 15:52:17 or perlguru Jul 25 16:06:55 DaKa2: My openslug native build got up to automake-native Jul 25 16:07:10 checking build system type... Invalid configuration `armv5teb-linux': machine `armv5teb' not recognized Jul 25 16:07:11 configure: error: /bin/bash lib/config.sub armv5teb-linux failed Jul 25 16:07:25 ah, did you read my notes in the wiki? Jul 25 16:07:30 but that's with the normal openslug.conf, so we need an openslug-native.conf Jul 25 16:07:33 ah. no :-) Jul 25 16:07:46 what's that URL again? Jul 25 16:07:48 http://www.nslu2-linux.org/wiki/HowTo/OpenSlugNativeCompileEnvironment Jul 25 16:07:56 "Using BitBake on the slug" Jul 25 16:08:07 that is a bit old, but has the main things Jul 25 16:08:21 I think the fix for your problem is: BUILD_ARCH = armeb Jul 25 16:08:49 but I dont let bb make any -native Jul 25 16:08:55 I ipkg install them Jul 25 16:09:09 ok, I'll create an openslug-native.conf tonight (12 hours away) unless someone else gets to it in the meantime. Jul 25 16:09:57 can we add the stuff from steps 3, 4, 5, 6 and 7 into openslug-native.bb ? Jul 25 16:09:58 Im working on crosscompilable perl with 100% test passed, Im down to just a few test failing Jul 25 16:11:15 rwhitby: yes, or.. you already put whats needed of 4 in there Jul 25 16:11:21 and 3 is there :-) Jul 25 16:11:49 yep. Do we need the forces on 6 any more? Jul 25 16:11:50 5 is there if you have the native sources in your ipkg source-list Jul 25 16:12:06 hm.. I dont know... have to check Jul 25 16:12:09 forgot about them.. Jul 25 16:12:20 they need to be built too Jul 25 16:12:23 I think Jul 25 16:12:23 yeah, 5 is covered currently, and will be replaced by your cross built perl Jul 25 16:12:55 in case I manage to make it 100% working from cross :) Jul 25 16:13:19 ok, the goal for today is to reduce that list down to 4 steps :-) Jul 25 16:13:36 Swap, user, openslug-native, master makefile. Jul 25 16:13:51 well, that should be possible Jul 25 16:14:04 dyoung-away: we need that cron job for monotone snapshot when you're able to do it ... Jul 25 16:14:25 back later Jul 25 16:41:48 uhm.. the python bb provides python, but it should really rprovides python, seems wrong.. Jul 25 16:43:37 and I don't get this ipkg-link and ipkg-utils thingy.. why didn't they just remove the ipkg-link binary from ipkg-utils so you can have them both.. oh well, some other day, bbl Jul 25 16:45:24 oh "Failed 3 test scripts out of 221, 98.64% okay.", 1 should get fixed by ipkg install procps, the other two are because of ccache was used and now it thinks that is CC == ccache on the slug too Jul 25 16:46:02 he he Jul 25 17:45:34 rwhitby, its ready... Jul 25 17:45:57 wget http://www.nslu2-linux.org/nslu2-linux.initial.db Jul 25 17:46:11 generaated once an hour with the private key stripped. Jul 25 17:46:18 (from nudi's main db) Jul 25 17:47:36 dyoung-web: can we set up a monotone.nslu2-linux.org vhost, and serve it from that? Maybe we should serve the Makefile from the monotone.nslu2-linux.org vhost too .... Jul 25 17:48:17 cause we will need the monotone.nslu2-linux.org/viewmtn on that vhost anyway ... Jul 25 17:49:22 I dunno, that mkight make it easier for viewmtn to work today.... Jul 25 17:49:23 so it will not be viewmtn.nslu2-linux.org? Jul 25 17:49:39 monotone.nslu2-linux.org/viewmtn . Jul 25 17:49:43 * ka6sox-office reconfigures. Jul 25 17:50:29 as soon as the new www.nslu2-linux.org is up, then monotone.nslu2-linux.org will point to it. Jul 25 17:50:31 Okay, when the vhost is ready, I'll adjust the monotone auto-stripping-copy scrip tto put in the right place. Jul 25 17:51:03 I didnt get to the makefile yet. Jul 25 19:14:15 hm, hostname.coreutils is dumb Jul 25 19:17:39 where's the real hostname command.. sigh Jul 25 19:18:28 busybox? Jul 25 19:18:56 DaKa2: can you reduce the priority on the coreutils hostname lower than the one in the image? Jul 25 19:19:03 no, net-tools Jul 25 19:20:00 but anyways.. the coreutils hostname makes no sense and breaks the init-script Jul 25 19:20:35 it doesnt support -F /etc/hostname Jul 25 19:55:19 do we have modprobe in the openslug image? Jul 25 19:55:29 its supposed to be there. Jul 25 19:55:46 ah, normal user doesn't have /sbin in path Jul 25 19:56:49 rwhitby@openslug:/lib/modules/2.6.12.2/kernel/drivers/usb/media$ sudo /sbin/insmod pwc.ko Jul 25 19:56:56 insmod: error inserting 'pwc.ko': -1 Unknown symbol in module Jul 25 19:57:03 :-( Jul 25 19:57:25 ah, video Jul 25 19:58:26 and while youre threre you might have to mknod /dev/video or something. Jul 25 20:02:03 pwc Logitech QuickCam 4000 Pro USB webcam detected. Jul 25 20:03:12 now what do I need for the microphone on it ? Jul 25 20:06:44 hmm - do we have usb audio enabled for openslug? Jul 25 20:13:05 I dunno, for the mic, youre in unchartered waters. Jul 25 20:18:15 I expect it just standard usb audio and I need to enable it in the defconfig ... Jul 25 20:19:13 dyoung-web: did you say you'd put w3cam or something in openslug for a test capture? Jul 25 20:20:05 yeah, its in there. Jul 25 20:20:14 or if its not, it got lost in the transfer. Jul 25 20:20:27 lookking Jul 25 20:20:45 yeah, its still in there. Jul 25 20:21:18 is it in openslug-packages? Jul 25 20:21:25 I have no idea. Jul 25 20:21:33 looking Jul 25 20:22:08 nope, so it doesn't exist yet for me :-) Jul 25 20:22:18 Nope not in there. Jul 25 20:22:40 ok, something to do tonight ... Jul 25 20:22:51 (for me to do ...) Jul 25 20:23:16 I wonder if we can hard-code kernel module dependencies somewhere ... Jul 25 20:23:29 (e.g. pwc depends on videodev, etc) Jul 25 20:23:51 yeah, I had that working for one of these camera drivers... Jul 25 20:23:57 did you run depmod? Jul 25 20:24:03 or did it get run rather Jul 25 20:25:38 ah, right - when I first did it it was via sudo, which didn't find depmod. you're right - depmod did it correctly the second time. Jul 25 20:25:54 but I meant kernel module ipkg dependencies Jul 25 20:26:32 back later Jul 25 21:23:50 rwhitby-web dyoung: I'm about to drop in the stuff which fixes the oe-symlinks (in monotone). Jul 25 21:24:12 cool. go fo rit Jul 25 21:24:18 Okay, I dont have anyhting pending to the Makefile, go for it Jul 25 21:24:22 This works for both my private (non-makefile) build environment and for the master makefile, on my system Jul 25 21:24:36 good enough for me :-) Jul 25 21:24:41 I'm pretty sure there's no dependencies there, but there may be teething problems. Jul 25 21:24:53 As Linus says "Testing is for Users" Jul 25 21:25:13 Well, its sure to work a whole lot better than it does right now (with berlios down again) Jul 25 21:25:27 Any serious problems just delete the include conf/distro/freeze.conf from openslug.conf (and unslung.conf if necessary) Jul 25 21:26:01 is freeze.conf generic for all OE distros ? Jul 25 21:26:29 yes Jul 25 21:26:58 The thing which may not work for some non-nslu2-linux guys is that I require all the bb files to be in one master directory (packages) Jul 25 21:27:01 should it be a bbclass instead? (not having seen it, this could be way off base question) Jul 25 21:27:10 I.e. packages/package/package.bb Jul 25 21:27:39 can't be a bbclass - it has to define BBFILES - that's pretty tricky :) Jul 25 21:28:30 back in a bit; then I can see if I can build anything again. Jul 25 21:28:38 It's in there now - changes to both org.nslu2-linux.dev (needs a bit more work, assuming it's all ok) and org.openembedded.nslu2-linux Jul 25 21:29:30 is it enabled automatically now? do I need to do anything for it to start working? Jul 25 21:29:46 make update-master then make should be enough Jul 25 21:30:16 (it does use PKGDIR if defined and ${TOPDIR}/openembedded if not - that worked fine when I tried it.) Jul 25 21:30:54 It isn't enabled on unslung because I have to have a complete to-the-end build before I can generate the unslung-packages.conf file. Jul 25 21:31:10 (And I see the glibc-2.2.5 failure in make unslung) Jul 25 21:32:07 jbowler-away: the comment at the top of freeze.conf needs updating Jul 25 21:32:12 (file name changes) Jul 25 21:33:14 package-dirs -> packages Jul 25 21:33:40 Ah, couldn't see it (yes, I changed the names ;-) Jul 25 21:34:44 so does local.conf need to be changed at all? or is the BBFILES value there just overridden ? Jul 25 21:35:03 It gets overridden. Jul 25 21:35:13 bb freeze builds new files. Jul 25 21:35:30 bb unfreeze empties those files and consequently reveals the original (symlinks) definition Jul 25 21:35:52 can you comment it out of local.conf.template, or change it back to openembedded in local.conf.template? Jul 25 21:35:58 Both those targets are strictly local - they don't put anything in the monotone (openembedded) tree. Jul 25 21:36:38 Yes: more changes are needed to the master makefile, but if there is something wrong with what I have done reversing it is currently trivial, once I start hacking the master makefile backing out the changes will be difficult. Jul 25 21:37:27 seems to be working here Jul 25 21:37:51 Instead of adding symlinks now just add (by hand) to openslug-packages.conf. When I have changed the master makefile 'unfreeze' will reveal all the packages (yes, all of them...) and freeze will generate the correct openslug-packages.conf Jul 25 21:38:33 can you just do a new bb and then refreeze without having to unfreeze? Jul 25 21:38:37 rwhitby-web: do you know of any workround for the glibc problem in unslung? Jul 25 21:38:59 jbowler-away: nope. I have no idea why it is happening, nor how to fix it. Jul 25 21:39:35 I do know that /home/unslung/firmware on nudi does *not* exhibit the problem (so can be used for comparison), and a rebuild of glibc in that dir (*without* updating source files) works. Jul 25 21:40:44 Oh, ok, I'll snapshot the work/* directories out of there then and generate an unslung-packages.conf Jul 25 21:41:43 In fact I suppose the symlinks are still ok for unslung - it's just that no one can build if they don't already have oe-symlinks (because of the svn problem). Jul 25 21:42:22 it would be good to move to unslung-packages now Jul 25 21:42:43 I want to get our builds away from berlios ASAP, they've screwed us one to many times. Jul 25 21:43:16 our trusted third party releases can stay there, but not anything that builds need. Jul 25 21:44:16 So far as I can see the OpenSlug-2.3-beta source tarball should build fine without svn. Jul 25 21:45:27 aparently there is a postfix package issue with it ... Jul 25 21:46:29 That's a build environment issue IIRC - make openslug-firmware-safe might work round it. Jul 25 21:48:52 is that the unset MFLAGS thingo? Jul 25 21:50:37 -safe cleans the environment for bitbake, so it removes MFLAGS - I can't remember whether that was the problem in this case. Jul 25 21:52:19 ok, so to add a new package now we just edit conf/distro/openslug-packages.conf and packages/meta/openslug-packages.conf Jul 25 21:52:43 any way those two can be combined? Jul 25 21:53:54 No - BBFILES happens too early on. I tried python on a simple list of packages (directories) and I discovered there was no bb.data Jul 25 21:54:00 jbowler-away: yes, unslung-packages.conf from /home/unslung/firmware on nudi should be a good starting point. Jul 25 21:54:50 I'm generating that now - I suspect I'll get a lot of 'missing bb files' errors because of old versions of things (like four different versions of ipkg in the work directory). Jul 25 21:55:40 and then we'll sever the line to berlios.de forever! Jul 25 21:57:22 What I said above was incorrect - bb unfreeze doesn't reveal the original package list, it reveals the openslug-packages.conf. Jul 25 21:57:46 I could change that to make it generate ${PKGDIR}/packages/*/*.bb Jul 25 22:04:04 conf/distro/unslung-packages.conf added... Jul 25 22:17:13 jbowler-away: if unfreeze doesn't generate */*.bb, then it will not be possible to do bb right? Jul 25 22:17:46 is there a bitbake command which just does one .bb file without worrying about dependencies? Jul 25 22:18:45 bitbake -b ..../foo.bb Jul 25 22:19:07 ok, so we can use that to add files to work and then freeze - no need to unfreeze. Jul 25 22:19:26 */*.bb makes it easier to add new packages reliably - otherwise it is necessary to find all the dependencies too. Jul 25 22:20:12 The problem is that there may be multiple-providers type problems and the frozen list (or the symlinks) can hide this. Jul 25 22:20:19 yeah, ok, that works. I can do it the quick way by not unfreezing, bb -b foo.bb, then refreezing, or the safe and slow way by unfreeze, bb foo, freeze. Jul 25 22:20:33 Yes Jul 25 22:20:43 */*.bb is better then Jul 25 22:20:56 Yes, and easy to do. I'll test it. Jul 25 22:21:17 at that point, feel free to remove all trace of oe-symlinks from everywhere/ Jul 25 22:21:26 I think I've found the glibc problem - I believe I broke it when I tuned the optimisations for openslug (it looks like something I put in there stops the glibc compile working on unslung). Jul 25 22:21:28 and we'll forget it ever happened ... Jul 25 22:22:18 ah - I didn't even think to diff the machine.conf files ... Jul 25 22:23:03 It looks like all the -f options get lost (including, I think, -fPIC which is added by the package build). Jul 25 22:24:51 if unslung works again, then I might be able to get the hat-trick tonight! Jul 25 22:25:44 FATAL: freeze: set $PKGDIR to the directory containing OE packages Jul 25 22:25:57 Do we need to set PKGDIR in local.conf.template? Jul 25 22:26:47 That's odd - it is set in conf/distro/freeze.conf Jul 25 22:28:05 Something strange is happening here - I do have it set in my environment, but not for a simple 'make unslung'. Jul 25 22:29:21 Hum... it works fine if I unset PKGDIR in my environment (i.e. bb freeze for unslung succeeds), and I don't have TOPDIR set. Jul 25 22:31:20 rwhitby@nudi:/home/slug/openslug$ echo $PKGDIR Jul 25 22:31:33 "/home/slug/openslug/openembedded/" Jul 25 22:31:39 (quotes added) Jul 25 22:31:59 set in setup-env Jul 25 22:32:09 it's not exported Jul 25 22:32:41 I.e. it is set, but it isn't in the environment - try running a sub-shell and checking for it. Jul 25 22:32:49 nod Jul 25 22:33:59 Maybe TOPDIR is somewhere different on your system. Jul 25 22:34:38 I don't define TOPDIR Jul 25 22:34:54 where is TOPDIR defined for you? Jul 25 22:35:19 Oops: bb.data.setVar('TOPDIR', os.getcwd(), data) Jul 25 22:36:35 That's fine inside the make. Jul 25 22:36:51 we have OEROOT defined in local.conf if that helps ... Jul 25 22:38:17 I believe exporting PKGDIR is best - that's what I do. Jul 25 22:41:36 jbowler-away: can you remove the two last remaining mentions of berlios.de from the master Makefile? Jul 25 22:42:11 Yes. I've got to fix setup-env too - I'll change it to export PKGDIR Jul 25 22:42:11 and feel free to change the Makefile to cause setup-env to export PKGDIR. Jul 25 22:42:21 snap Jul 25 22:43:59 Any idea why PKGDIR is on BBPATH? Jul 25 22:44:09 Oh, ok, beause of conf, not packages. Jul 25 22:44:16 can you also make the unslung Makefile include unset MFLAGS and unset MAKEFLAGS in setup-env creation like the openslug Makefile does? Jul 25 22:44:43 also remove all traces of oe-symlinks from the unslung and openslug Makefiles :-) Jul 25 23:03:47 back later Jul 25 23:22:03 jbowler-away: I pushed the Makefile changes Jul 25 23:22:57 Okay, what is the protocol for adding stuff to the foo-conf file? Manual editing? Jul 25 23:23:35 new one gets created by the Makefile - we'll have to manually edit or delete and recreate Jul 25 23:24:03 Which Makefile changes? Jul 25 23:26:17 oe-symlinks and setup-env Jul 25 23:26:39 Ah, I was just doing those. Jul 25 23:27:04 I realised I could do it on nudi as nslu2-linux ... Jul 25 23:27:52 Actually, you cant. Jul 25 23:28:28 because when I copy that database and strip the key it will lose the public key too. Jul 25 23:28:57 if you want to be able to do edits as nslu2-linux, I'll need to restructure some things to pull a private db. Jul 25 23:29:38 its totally doable. I was being lazy by using the existing db. Jul 25 23:31:14 I just did, and have done so before, so it's too late :-) Jul 25 23:31:37 wait, you have before? Jul 25 23:31:47 okay, it will probably work then. Jul 25 23:32:05 maybe the before was before the recreation of the db Jul 25 23:32:05 I guess we'll find out in 37min. Jul 25 23:32:14 but I will need to again, e.g. from treo Jul 25 23:35:22 It doesn't matter losing the public keys from the list, they will come back. Jul 25 23:40:04 rwhitby-web: I have a better fix for the PKGDIR problem by setting it in local.conf (from OEROOT)... just testing it. Jul 25 23:42:27 jbowler-away: cool Jul 25 23:42:37 BTW, glibc build confirmed on nudi for me :-) Jul 25 23:44:27 jbowler-away: Ahh, so I can ignore the warning messages about the "missing key ccalbhalbhlah" type messages and they will re-appear during a subsequent pull or something? Jul 25 23:46:20 Umm, no, maybe not - I've done dropkey on my signing key (including the public bit) and still had a database which worked, but I did pull from another db with the key almost immediately. Jul 25 23:47:12 the database works, it just complains about missing keys. I havnt tried updating that database to see if keys get re-put though.. Jul 25 23:49:28 It seems like it might be a problem. **** ENDING LOGGING AT Mon Jul 25 23:59:56 2005