**** BEGIN LOGGING AT Thu Jun 30 23:59:57 2005 Jul 01 00:09:21 I propose moving strace to b0rken packages (jacques asserted earlier that no version currently in OE would work anyway). Jul 01 00:38:50 With strace moved openslug-packages (uclibc) completes fine. Jul 01 00:47:36 any idea on the depmod problem? Jul 01 00:54:23 Yes: the unslung kernel.bb in packages/linux has the linksys version number - so it gets "23". Jul 01 00:54:44 Need to understand what changed/why - do you have the viewmtn URL? Jul 01 00:54:59 Hum, probably need to look in BK... Jul 01 00:55:51 ah, so we need to change the PV for our kernel? Jul 01 00:56:12 makes sense Jul 01 00:56:22 That would fix it, but I don't understand why they use PV for depmod despite defining KERNEL_MAJOR_VERSION correctly later on. Jul 01 00:56:38 (The latter comes from the linux header file) Jul 01 00:58:17 maybe we can fix that Jul 01 00:58:20 pb added it on June 28. Jul 01 00:59:12 pb joined #oe half n hour ago ... Jul 01 00:59:20 heh :-) Jul 01 01:03:23 I suspect it's just an accident. Jul 01 01:04:41 good catch anyway Jul 01 01:09:11 ok, so we change the version of unslung kernel.bb? Jul 01 01:09:31 2.4.xxx-2.3r29? Jul 01 01:09:37 That's the easiest thing to do. Jul 01 01:09:47 let's do it Jul 01 01:09:48 Let me look at the regexp. Jul 01 01:10:12 back later ... Jul 01 01:15:38 2.4.22.l2.3r29 Jul 01 01:16:28 Because, while portmap does it, nothing else has a hyphen in PV - I'm worried something might break if one is there. Jul 01 01:18:43 Ok, testing it, it's just a rename of the two unslung-kernel .bb files, and it certainly fixes the depmod problem. Jul 01 01:18:57 sleep... Jul 01 01:41:48 wohoo Jul 01 01:42:14 suddenly, monotone starts stuff to resolve a three-way merge Jul 01 04:22:58 NAiL: ping Jul 01 04:23:59 pong Jul 01 04:24:35 NAiL: you told something that udev is not working if the system is booting from a disk Jul 01 04:24:45 yeah Jul 01 04:24:52 It "kindaworks" Jul 01 04:24:55 i cannot confirm that, it's working here Jul 01 04:25:07 But it prevents you from unmounting /initrd Jul 01 04:26:20 I aim to add udev at a later stage. After I've seen exactly what happens at boot time Jul 01 04:26:33 I'm going to solder serial on my slug in a short while Jul 01 04:26:44 I just need to go out and get a solder sucker Jul 01 04:26:51 yeah, let's get a stable openslug build in monotone before adding udev to it Jul 01 04:26:54 NAiL: i got the chip but i need to build up a board now Jul 01 04:26:59 rwhitby: Working on it. Jul 01 04:27:11 mr_claus: what chip? A max3232? Jul 01 04:27:11 NAiL: heat it up, then bang it on the table :-) Jul 01 04:27:18 NAiL: yes Jul 01 04:27:26 rwhitby: hahaha. Good idea ;-) Jul 01 04:27:48 that's my patented percussive solder removal tool :-) Jul 01 04:28:06 So it works? :) Jul 01 04:28:13 just don't bang too hard and crack the PCB .... Jul 01 04:28:42 another one is to heat it up and then stick a cold needle through it Jul 01 04:28:51 cold needle? Jul 01 04:29:12 I tried banging it, but I guess I'm not banging hard/fast enough. Jul 01 04:29:31 do the middle two first. the outside two are the hardest Jul 01 04:29:44 if youre lazy, you only need to do 3 holes. Jul 01 04:29:56 because you can heat the 4th while youre shoving the header in. Jul 01 04:29:56 I'll be damned Jul 01 04:30:01 another way is to clear the middle two, then put the header on and push it through while you heat the outside two, rocking it back and forth Jul 01 04:30:16 One hole clear : Jul 01 04:30:30 Go NAiL, Go NAiL, Go NAiL! Jul 01 04:30:35 twio Jul 01 04:30:43 which method? Jul 01 04:31:14 bang Jul 01 04:31:44 middle two? Jul 01 04:32:12 yeah Jul 01 04:33:14 Can't clear out pin4 though Jul 01 04:33:45 yeah, it's connected to a middle ground/power plane inside the PCB. that sucks the heat right away Jul 01 04:34:04 do the shove Jul 01 04:35:01 well.. *hopefully* I don't need pin1, right? Jul 01 04:35:19 Got two usb-serials Jul 01 04:35:28 not if the pl2303 is powered from usb and does 3.3V correctly Jul 01 04:35:35 bbiab Jul 01 04:47:32 I've been using a a combination of the precussive removal and wick. Jul 01 04:49:15 ok, serial header attached Jul 01 04:50:40 *drum roll* Jul 01 04:51:15 now, where did I put that paper that said which pin was which on this cable? Jul 01 04:51:58 white was ground? Jul 01 04:52:41 ok, I'll try with that cable instead Jul 01 04:54:35 +!!! Jul 01 04:54:48 I got *something* Jul 01 04:55:04 http://groups.yahoo.com/group/nslu2-linux/database?method=reportRows&tbl=11 Jul 01 04:55:08 oops Jul 01 04:55:17 neat-o Jul 01 04:55:20 It works! Jul 01 04:55:25 * NAiL goes bananas Jul 01 04:55:40 * dyoung eats a banana. Jul 01 04:55:56 It even has a nice blue light that flashes every time something passes over the serial line ;-) Jul 01 04:56:11 s/flashes/lights up the entire room/ Jul 01 04:56:25 Electrical tape is your friend. Jul 01 04:56:53 Didn't think there was much output from the original firmware? I got like.. heaps. Jul 01 04:57:02 rwhitby logs into a banana Jul 01 04:57:18 I *am* logged into banana Jul 01 04:57:53 dyoung: Blue was TX, White was GND, Black was RX. Jul 01 04:58:05 I only remember white was gnd. Jul 01 04:58:25 I figure thats enough that you can figure out the rest. :-) Jul 01 04:58:30 hehe, yeah Jul 01 04:58:43 blue was RX, even... connected to the TX pin Jul 01 04:58:59 Every time I touched the blue wire, minicom went crazy :-P Jul 01 04:59:45 Ok, so now I have an OC'ed, serialized slug. *Now* I'm happy Jul 01 05:00:47 OC? Jul 01 05:00:54 Oh, overclock Jul 01 05:01:05 OC means something else to me. Jul 01 05:03:53 of course? Jul 01 05:07:01 NAiL: it seems the pivot_root could not be finished because some files are open at the old rootfs Jul 01 05:08:22 mr_claus: You got serial? Jul 01 05:08:44 NAiL: not but there is no other reason i think Jul 01 05:08:57 NAiL: i check the open files now Jul 01 05:09:02 I'll be checking that very soon ;) Jul 01 05:10:11 sh 931 root cwd DIR 31,4 0 1 /initrd/ Jul 01 05:10:13 Open Collector. Jul 01 05:10:48 hmm, no that was me Jul 01 05:12:03 exec <&- >&- 2>&- Jul 01 05:12:06 what is that doing? Jul 01 05:12:23 that makes sure stdin/stdout is redirected Jul 01 05:12:56 If I get it correctly, it closes the filedescriptor on /dev of the flash device, and opens it on /dev on the root disk. Jul 01 05:13:05 (Anyone care to confirm? =) Jul 01 05:13:33 that's what it's supposed to do Jul 01 05:13:44 yeah Jul 01 05:13:57 Hmm.. what should I call my new slug? Jul 01 05:14:32 Don't call it banana, cause the OpenSlug build slug is called that Jul 01 05:14:44 then use orange Jul 01 05:14:53 agent orange ;) Jul 01 05:15:09 strawberry Jul 01 05:15:54 heh.. the musicvideo for hollaback girl is on the TV. They just flashed "banana" Jul 01 05:17:27 gunslinger ;) Jul 01 05:19:56 btw, is there any way to build upslug with the current repo? Jul 01 05:22:44 thath would be a great addition. Jul 01 05:23:31 Thanks for volunteering! Jul 01 05:24:17 ./openembedded/packages/upslug Jul 01 05:24:22 hmm.. and that does what? Jul 01 05:25:04 doesnt that make a arm binary of upslug? Jul 01 05:25:13 or does it make the native? Jul 01 05:25:16 * dyoung dunnos Jul 01 05:26:43 It's jbowlers work Jul 01 05:27:12 But I get "nothing provides"... Jul 01 05:28:07 is it in the symlnks? Jul 01 05:28:21 yup Jul 01 05:39:53 Okay now that homeslug is Turbo'd and I have a shiny new 512MB stick on it, which OS to load on it... Jul 01 05:40:30 Windows CE Jul 01 05:40:39 * NAiL hides Jul 01 05:40:49 OpenSlug 1.12 :-) Jul 01 05:41:02 *sigh* Jul 01 05:41:11 I need another slug for unslung :P Jul 01 05:41:30 Buying one next month I guess Jul 01 05:42:02 argll, udev is locking anything on initrd Jul 01 05:42:11 yes Jul 01 05:42:32 but udev is started after the chroot Jul 01 05:42:34 which it shouldn't do, since it's started after the pivot_root Jul 01 05:42:59 the pivot_root is not unmouning /initrd Jul 01 05:43:28 it only transfer processes to the new rootfs Jul 01 05:43:50 can anybody confirm if that is right? Jul 01 05:45:44 that is correct Jul 01 05:46:10 we specifically don't umount /initrd, but we do expect that no processes are using it, so it can be umounted by the user Jul 01 05:46:44 Any reason why initrd isn't unmounted? Jul 01 05:46:45 ok, so i could try (only for testing) to unmount it after the chroot is done Jul 01 05:47:06 NAiL: but udev should not use the initrd fs Jul 01 05:47:15 mr_claus: No, it shouldn't. Jul 01 05:49:08 udevd 686 root txt REG 31,4 26872 510 /initrd/sbin/udevd Jul 01 05:49:18 udevd was started on the old rootfs Jul 01 05:49:36 there are a lot of entries which are used on initrd Jul 01 05:49:36 how? Jul 01 05:50:00 i dont know yet, the libc of /initrd is opened by udev too Jul 01 05:50:26 perhaps the startup script is buggy Jul 01 05:50:35 of udev, but i have to look into Jul 01 05:50:55 upgrading new slug Jul 01 05:52:39 I put Unslung 5.5 on it, but I think I'm gonna go with openslug-1.12 . Jul 01 05:53:13 Why 1.12? Jul 01 05:53:21 mostly because it works. Jul 01 05:53:56 unless someone has figured out why openssh is flaky on 2.x Jul 01 05:54:10 oh.. no. Jul 01 05:54:24 Using dropbear. Haven't tried openssh Jul 01 05:54:59 Does it have anything to do with the changes to /dev/random? (Or whatever random-device it uses) Jul 01 05:55:42 * dyoung-zzzz dunnos Jul 01 06:03:16 New slug openslugged. Jul 01 06:10:52 Hmm... I can't login through the serial console Jul 01 06:12:29 removed the startup links but udev is starting again while booting Jul 01 06:13:12 but.. from the new rootfs Jul 01 06:13:27 aha Jul 01 06:13:29 it seems that the hotplugd is the reason why udev is starting Jul 01 06:13:42 /etc/hotplug.d/default/10-udev.hotplug Jul 01 06:14:17 well.. I'm gonna run off and see this girl. Probably be back later sometime ;-) Jul 01 06:14:29 :) Jul 01 06:17:02 dyoung-zzzz: heh... I was wrong. White wasn't GND. It was TX. Black was GND ;) Jul 01 06:17:10 Now I can login through the serial console :P Jul 01 06:19:29 btw, anyone remember how I specify which directory the ipkgs go to? I'd like to deploy them directly into a ftp-accessible directory Jul 01 06:28:23 [g2]:) Jul 01 06:28:30 *bzzzt* Jul 01 06:28:45 <[g2]> :) Jul 01 06:28:49 <[g2]> nothing like a little jolt in the moring! Jul 01 06:28:57 Got my new slug today. deunderclocked in less than 13 minutes :) Jul 01 06:29:13 (From the mailman came to the door) Jul 01 06:29:29 <[g2]> you seem to be having a bunch of fun Jul 01 06:29:30 And I got serial working on it as well :D Jul 01 06:29:44 <[g2]> cool! Jul 01 06:29:46 <[g2]> congrats Jul 01 06:29:58 <[g2]> did you figure out the other serial issue ? Jul 01 06:30:12 I need to get a bigger resistor, I guess Jul 01 06:30:32 <[g2]> you've got an extra 10K right ? :) Jul 01 06:30:50 haha, yeah. But my tip is way bigger than those two 10K's ;) Jul 01 06:31:03 <[g2]> bragging again Jul 01 06:31:14 hehe Jul 01 06:31:23 hmm... the one I left on the table is gone Jul 01 06:31:42 I've probably put something on top of it :-P Jul 01 06:33:04 Now I've gotta get a new power brick... Jul 01 06:33:37 Check the OverClock howto, at the bottom, to see how much power the slug draws in total. Jul 01 06:34:07 If I'm gonna connect two 2.5" drives, I'm gonna need more power. Jul 01 06:36:20 btw, my serial is cooler than yours. Mine has activity leds! (Blue for RX, red for TX) Jul 01 06:38:44 <[g2]> oooooooh Jul 01 06:39:25 I can see how far it is in the bootprocess without opening the console! ;) Jul 01 06:39:37 (Not if it stops though) Jul 01 06:40:45 * NAiL wanders off Jul 01 06:40:49 later! Jul 01 06:58:24 anybody know how the programs in /etc/hotplug.d/ will be called? Jul 01 07:01:15 <[g2]> a little Jul 01 07:01:42 there is a file installed at /etc/hotplug.d/default/10-udev.hotplug Jul 01 07:01:53 when that program will be started at bootup? Jul 01 07:02:14 it seems there is no startscript, it's started before chroot step perhaps Jul 01 07:02:54 <[g2]> that makes some sense Jul 01 07:03:03 <[g2]> the starting before chroot part Jul 01 07:03:33 not really because udev is locking a lot of files in /initrd then and /initrd cannot be unmounted Jul 01 07:03:38 <[g2]> it's hard for one to chroot if the the usb disk isn't recognized Jul 01 07:04:10 hotplug is ok, but udev needs to be started later after the chroot step Jul 01 07:04:27 <[g2]> and the hotplug event, acutally coldplug at boot gets that setup Jul 01 07:05:22 <[g2]> I'm assuming you are talking about OpenSlug which is actually using hotplug-ng Jul 01 07:05:30 yes Jul 01 07:05:32 <[g2]> the C programs versus the scripts Jul 01 07:08:14 <[g2]> It may not be this way, but I think the mount -t foo causes the foo module (and it's dependent modules) to be loaded Jul 01 07:08:33 <[g2]> and I think they trigger the hotplug event Jul 01 07:09:17 <[g2]> either that or there a something in /etc/modules.??? Jul 01 07:10:14 i think it could be the -t fs stuff Jul 01 07:10:41 but an easy way could be to take out the udev from the hotplug structure Jul 01 07:11:05 udev will wie started in /etc/rcS.d/S04udev Jul 01 07:12:24 <[g2]> I haven't played with udev Jul 01 07:37:50 yawn. Jul 01 07:40:34 hey jacques Jul 01 07:41:08 check out this branching: http://monotone.vanille.de/viewmtn/revision.psp?id=a03f0834b5caae5065d7445ef3fed2c8700023a8&ancestry_limit=0 Jul 01 07:42:14 night all Jul 01 07:42:32 <[g2]> nite Jul 01 07:42:37 nite nite Jul 01 07:42:38 <[g2]> looks cool Jul 01 07:44:53 man that is too cool Jul 01 07:45:06 soooo.... Jul 01 07:45:12 is anyone here going to ols? Jul 01 07:47:54 <[g2]> giel, I don't think I'll be going this year Jul 01 07:48:00 <[g2]> how are you doing ? Jul 01 07:48:10 I'm registered but probably not going. Jul 01 07:48:19 heh Jul 01 07:48:23 verrreh buzzeh Jul 01 07:48:30 i have to, i'm a speaker ;) Jul 01 07:48:50 <[g2]> If jacques is registered I might go a him :) Jul 01 07:49:05 <[g2]> go as he Jul 01 07:49:12 <[g2]> go as him Jul 01 07:49:14 so anyway, it might just so happen that i'll fire up my slug somewhere next week and try the latest openslug Jul 01 07:49:24 so, that's when i remembered you guys ;) Jul 01 07:49:32 giel: got your soldering iron handy? Jul 01 07:49:43 more or less, why? Jul 01 07:49:51 <[g2]> Ooohhh Jul 01 07:50:38 <[g2]> http://www.nslu2-linux.org/wiki/HowTo/OverClockTheSlug Jul 01 07:51:04 ooh! Jul 01 07:51:55 w00t! will do that! Jul 01 07:52:09 but i think i first have to have a booting slug, to see if i'm succesfull :) Jul 01 07:52:25 there's some old openslug mess on it now from a couple of months ago Jul 01 07:56:40 so, reading up on the status of openslug, you've come a long way since i sneaked out Jul 01 07:57:02 <[g2]> there's a new openslug co-lead Jul 01 07:57:12 <[g2]> jbowler is a real ass kicker Jul 01 07:57:42 <[g2]> he's done a lot of great work Jul 01 07:57:45 apex! Jul 01 07:57:50 it's working now? Jul 01 07:57:59 <[g2]> been working for a month or two Jul 01 07:58:12 * giel has been away for too long Jul 01 07:58:18 <[g2]> I'm booting from the kernel *inside* the jffs2 partition :) Jul 01 07:58:21 <[g2]> with APEX Jul 01 07:58:25 uhuh Jul 01 07:58:29 that's what i wanted all along Jul 01 07:58:38 <[g2]> 1.2.17 Jul 01 07:58:54 <[g2]> it's slower on boot Jul 01 07:59:00 <[g2]> obviously Jul 01 07:59:20 ok, priority setting: 1) get a working slug by installing latest openslug 2) do the clock mod 3) try apex Jul 01 07:59:32 <[g2]> cool Jul 01 07:59:40 <[g2]> you're JTAG enabled just in case ? Jul 01 07:59:57 hey folks Jul 01 08:00:06 no, but i know how to do it and have access to all kinds of jtag equipment at work Jul 01 08:00:15 so, if i brick it, i'll most pbly be able to fix it Jul 01 08:00:49 oh, and my arm reference manual got it, so now i can actually understand arm assembly ;) Jul 01 08:00:54 <[g2]> Ok you know what you're getting into Jul 01 08:01:19 <[g2]> I've run APEX for months without issue though Jul 01 08:01:31 yeah, i was away because i was busy working with embedded devices at work :) Jul 01 08:01:42 <[g2]> cool Jul 01 08:01:46 <[g2]> embedded linux ? Jul 01 08:01:49 and what's the worst that could happen? a bricked slug... Jul 01 08:02:09 <[g2]> nod. It's not like your hair catching fire Jul 01 08:02:31 dun have any Jul 01 08:02:34 atm Jul 01 08:02:48 sorry to interrupt, but I had 2 quick questions Jul 01 08:02:53 shoot Jul 01 08:03:01 1. is it just me, or has ls disappeared from coreutils? Jul 01 08:03:19 2. I've been working on getting openvpn working Jul 01 08:03:25 but I'm new to package development Jul 01 08:03:30 and I built the ipk Jul 01 08:03:33 and installed Jul 01 08:03:42 and the binary dumped in /opt/sbin Jul 01 08:03:58 has rx permissions only for root Jul 01 08:04:19 which I don't think should be the case Jul 01 08:04:27 but I'm not sure where to adjust that Jul 01 08:04:35 well, if it's in sbin then the permissions are okay Jul 01 08:04:41 question is, should it be in sbin Jul 01 08:04:53 ok Jul 01 08:04:55 I would argue no Jul 01 08:05:03 since it's a combined server/client binary Jul 01 08:05:23 I'm not sure why it ended up there Jul 01 08:05:25 but openvpn adds network devices as a client? or doesn't it Jul 01 08:05:26 ? Jul 01 08:05:40 I think it does Jul 01 08:05:52 sounds to me that's a task typically done by root Jul 01 08:06:05 but i'm just guessing here, i don't know openvpn :) Jul 01 08:06:39 anyway, i'm off for the weekend, catch you guys on monday Jul 01 08:06:41 hm Jul 01 08:06:51 bye Jul 01 08:08:59 hm Jul 01 08:09:08 if I do an ipkg install openvpn Jul 01 08:09:21 <[g2]> giel, have a good weekend Jul 01 08:09:25 <[g2]> thx for stopping by Jul 01 08:09:37 it goes to /opt/sbin with permissions 775 Jul 01 08:09:58 but if I install from the local ipkg I created Jul 01 08:10:03 it goes in with 700 Jul 01 08:10:17 is this standard? Jul 01 08:17:57 mr_claus, NAiL: if you don't want initrd mounted after boot rmdir it. Jul 01 08:18:09 (I.e. rmdir it on the final root partition) Jul 01 08:28:02 jbowler-zzz: it seems that udev is started to early and lock files on the /initrd fs Jul 01 08:28:26 jbowler-zzz: i think udev should start later from the final rootfs Jul 01 08:28:47 jbowler-zzz: then all things will be handled like without udev Jul 01 08:31:25 mr_claus: you've discovered, I believe, that the kernel is starting it. Jul 01 08:31:46 mr_claus: it is required to have the devices correct, so kill it after the streams are closed in the switch function. Jul 01 08:32:06 (The shell syntax >&- closes a file descriptor - 1 in that case). Jul 01 08:32:28 jbowler-zzz: if removed the file /etc/hotplug.d/default/10-udev.hotplug and it was not started Jul 01 08:33:14 Yes, but you can't do that, it will disable udev on a flash boot - that's the recovery mechanism. Jul 01 08:34:24 i know but if i remove the file and it isn't started then the kernel is not starting udev or do you mean the kernel because hotplugd is starting udev? Jul 01 08:35:08 Could you paraphrase that question? Jul 01 08:36:05 is hotplugd part of kernel? Jul 01 08:36:26 Yes - there kernel directly invokes /sbin/hotplug Jul 01 08:36:42 (iirc - not sure about the exact path - it's in the kernel sources) Jul 01 08:37:08 Why does udev require an init script if it is hotplugged? Jul 01 08:37:31 i dont know, the script comes with the package Jul 01 08:38:16 It sounds like one or the other should be there (init script or hotplug script). I'd be inclined to use hotplug and make sure the daemon is killed after the pivot_root. Jul 01 08:39:42 The reason for that preference is that the file system modules are modprobed from the kernel, and they require the devices - so if the kernel guys want to do this let them do everything... Jul 01 08:39:54 would it be started new by hotplugd or do i need to start it manually again? Jul 01 08:40:36 so that means kill udev before chroot and start it after chroot has finished? Jul 01 08:41:24 It should not be necessary to restart it (but I might be wrong), and it should be killed where the streams are closed - probably just before. Jul 01 08:44:30 Maybe part of the problem here is that we have some big set of init scripts then are trying to force hotplug on top without rewriting the init scripts. Jul 01 08:44:38 ACTION  Jul 01 08:46:42 so you mean i should kill it just before swivel is called? Jul 01 08:47:22 swivel is changing to new-root and is calling pivot_root Jul 01 08:47:47 No: I mean you should kill it immediately before the streams are closed. Jul 01 08:48:25 swivel() { Jul 01 08:48:25 cd "$1" Jul 01 08:48:25 exec <&- >&- 2>&- Jul 01 08:48:32 (I said after before because the streams use a device, but I think killing it before may make it easier to debug...) Jul 01 08:48:50 so before swivel() is called would be ok Jul 01 08:50:11 By 'immediately before' I mean the preceding command. Jul 01 08:50:40 so exactly before the exec? Jul 01 08:50:47 This is just my opinion, but you haven't offered any argument for me to change it. Jul 01 08:51:11 There are several exec's in there. You quoted the lines I'm talking about. Jul 01 08:51:15 i would do that, it's no problem, i will change the funtions file for testing Jul 01 08:56:30 hmm, sometimes, if i write reboot at the console it doesn't come up again Jul 01 08:57:40 The warm (branch to 0) reset never works, the hard reset is intermittent due to a bug on the chip. Jul 01 08:59:02 it works sometimes and sometimes not Jul 01 09:01:47 hmm, the killall udevd didn't work or it was restarted automaticly before the pivot_root was done Jul 01 09:02:46 That would seem to be an excellent justification for doing it after the pivot_root. Jul 01 09:03:44 hrhr, i will insert it a couple of time and write process lists to see whats happening :) Jul 01 09:05:22 I expect there is some official way of killing udevd. Also killall needs to read the process list, and /proc is probably umounted. Jul 01 09:06:33 <[g2]> jacques, you know (playing Monday morning QB) that article should have been titled "Adventures in Normal Clocking" Jul 01 09:08:09 jbowler-zzz: yeah, proc is unmounted it seems, there is no processlist Jul 01 09:11:44 [g2], yes I agree Jul 01 09:12:06 I forgot the hysteria that comes with the mention of overclocking among the ignorant Jul 01 09:14:31 jbowler-zzz: the Makefile of udev is using "- killall $(DAEMON)" Jul 01 09:19:57 <[g2]> :) Jul 01 09:30:19 <[g2]> anyone familiar with apache2 php5 config ? Jul 01 09:54:16 does anybody know a software which is using dyndns on client and server? Jul 01 11:09:18 anyone in here care to test libusb on unslung? Jul 01 11:09:31 guess everyone is sleeping... Jul 01 11:11:57 hi DaKa2 Jul 01 11:14:22 hello :) Jul 01 11:15:57 did you fix libusb? Jul 01 11:16:15 I think so... do you have time to test it? Jul 01 11:16:32 DaKa2: sluggers never sleep. There's always someone awake :D Jul 01 11:16:45 well both of my slugs are openslug :-( Jul 01 11:16:48 :)) Jul 01 11:16:55 ahh, dammit.. mine is too.. Jul 01 11:17:07 hmm, both mine are as well Jul 01 11:17:15 what does libusb do anyway? Jul 01 11:17:43 well... only way for userspace to talk directly to usb in 2.6 Jul 01 11:18:02 or, atleast the only good way Jul 01 11:18:13 aha Jul 01 11:18:18 sane and other things use Jul 01 11:18:22 it Jul 01 11:19:25 ah, ok Jul 01 11:19:27 i tried getting gphoto2 and capture working on my slugs with no success Jul 01 11:19:33 they both use libusb Jul 01 11:19:41 Which means I'll need it to get my Canon cam working. Jul 01 11:19:41 well, it should work with some fixed packages Jul 01 11:19:48 maybe they will work now Jul 01 11:19:48 http://david.thg.se/saker/libusb/ Jul 01 11:20:27 It builds & works on openslug? Jul 01 11:20:28 DaKa2, which file has the patches? Jul 01 11:20:34 yup Jul 01 11:20:41 neat-o Jul 01 11:20:49 jacques: patches in unslung-dir Jul 01 11:20:59 DaKa2, ah ok Jul 01 11:21:06 i was looking in openslug Jul 01 11:21:23 sane works and can scan, but the scanned image is a bit, well, strange Jul 01 11:21:29 probably some bug in sane Jul 01 11:21:44 You don't have permission to access /saker/libusb/unslung/debian-changes.patch on this server. Jul 01 11:21:49 lol Jul 01 11:21:54 wait a sec Jul 01 11:22:12 there Jul 01 11:22:26 ok got it Jul 01 11:22:30 is that the only change ? Jul 01 11:22:49 yup Jul 01 11:22:56 cool, thanks Jul 01 11:23:33 the openslug package is built from the debian source, but that patch is the only important stuff that differs Jul 01 11:24:44 just wish someone could test the unslung one, dont feel like reflashing my perfectly working openslug Jul 01 11:25:01 jacques: gphoto2 builds? Jul 01 11:25:08 DaKa2, lol neither do I :-) Jul 01 11:25:37 NAiL, yeah but it segfaults Jul 01 11:25:56 humm Jul 01 11:25:59 not that good :P Jul 01 11:26:13 tried building it natively? Jul 01 11:26:21 that's the only way I built it Jul 01 11:26:43 oh Jul 01 11:27:17 it could be beacuse of broken libusb and it might work now, will try later Jul 01 11:27:48 <[g2]> DaKa2, that's a nice quoute "my perfectly working openslug" :) Jul 01 11:28:06 hehe, well, it is :) Jul 01 11:28:23 <[g2]> OpenSlug 2.0 ? Jul 01 11:28:29 yup Jul 01 11:28:38 <[g2]> binary or source ? Jul 01 11:28:44 source Jul 01 11:28:53 <[g2]> did you do the native compile ? Jul 01 11:29:20 i have compiled a few things nativly yes Jul 01 11:29:50 <[g2]> did you roll your own compile environ or use the HowTO from the wiki ? Jul 01 11:30:04 I used the howto, worked really nice Jul 01 11:30:17 <[g2]> excellet Jul 01 11:30:21 <[g2]> excellent Jul 01 11:30:25 only needed to make binutils Jul 01 11:30:35 [g2]: I've set up the dev env like three times. It's worked every time :D Jul 01 11:30:38 any reason why its not in the archive? Jul 01 11:30:38 Not tempted by the turbo mod? :) Jul 01 11:30:52 <[g2]> it is in the BK archive :) Jul 01 11:31:00 mine is running at 266,24 bogomips now :-) Jul 01 11:31:07 <[g2]> cool Jul 01 11:31:10 :) Jul 01 11:31:37 <[g2]> I'm wait for the monotone stuff to settle down a bit Jul 01 11:31:52 me too Jul 01 11:31:53 well, is there a .ipk for binutils? I made mine from the .bb-file from BK Jul 01 11:32:01 <[g2]> then I'll put the changes in for the script or just create the meta package for the dev environ Jul 01 11:32:17 DaKa2: Now there is, if you pull from monotone. Jul 01 11:32:21 gnu-config is built as well Jul 01 11:32:29 nice Jul 01 11:32:36 <[g2]> nod to both of you Jul 01 11:33:09 <[g2]> changes in OE from a week ago added glib 2.3.5 and a fix to include the glibc-linux headers automatically Jul 01 11:33:19 <[g2]> that's what I've been testing/running with for the last week Jul 01 11:33:51 It's working out? Jul 01 11:33:58 <[g2]> all is fine and so if monotone is sync'ed with the OE changes it should work with just our script Jul 01 11:34:02 <[g2]> no tarball needed Jul 01 11:34:24 and no ln -s'ing ? Jul 01 11:34:30 <[g2]> I cut and pasted that script in openslug-init and just ran it Jul 01 11:36:00 what the? Jul 01 11:36:02 grep: /root/nslu2/openslug/tmp/cross/armeb-linux/lib/libstdc++.la: No such file or directory Jul 01 11:36:05 sed: /root/nslu2/openslug/tmp/cross/armeb-linux/lib/libstdc++.la: No such file or directory Jul 01 11:36:08 libtool: link: `/root/nslu2/openslug/tmp/cross/armeb-linux/lib/libstdc++.la' is not a valid libtool archive Jul 01 11:36:32 That is when I'm building on the slug. Where on *earth* does it get that path from? That's a path from my dev box. Jul 01 11:37:04 I remember getting something like that too, I just symlinked it, but ther probably is something better to do Jul 01 11:37:09 Now that we're talking native dev environment anyhoo ;) Jul 01 11:39:20 the path comes from a .la file Jul 01 11:39:33 * NAiL does a recursive grep Jul 01 11:39:42 some ppl say they should be fixed, personally I just remove them Jul 01 11:40:05 fscking libtool Jul 01 11:40:17 I'd say they should be changed ;) Jul 01 11:42:29 I'd say you should fix it in the fastest and ugliest way Jul 01 11:42:46 that's rm :-) Jul 01 11:43:13 rm /usr/lib/libstdc++.la? Jul 01 11:43:39 exactly, rm what? Jul 01 11:43:52 NAiL, is the bad path in that .la ? Jul 01 11:44:15 yea Jul 01 11:44:29 It continues to compile now Jul 01 11:44:38 now that you did what? Jul 01 11:44:48 i mv'd the file, to be on the safe side Jul 01 11:44:51 heh Jul 01 11:45:09 wont you need libstdc++.la for compiling c++ stuff? Jul 01 11:46:13 no Jul 01 11:46:36 hm.. ok Jul 01 11:46:38 What's it there for? Jul 01 11:46:52 it's a libtool thing Jul 01 11:47:10 AFAICT they have no benefit (tho I'm sure they are supposed to) Jul 01 11:47:10 find / -name "*.la"|xargs rm -rf Jul 01 11:47:19 ;) Jul 01 11:47:48 ahh, right.. it's the .a ones you need.. Jul 01 11:48:08 even those are only for static linking Jul 01 11:48:44 yup.. thats what I thought the .la was, Jul 01 11:49:08 guess I should go back to school... Jul 01 11:49:58 libtool libraries are .la Jul 01 11:50:17 .la are text files Jul 01 11:50:44 # Please DO NOT delete this file! Jul 01 11:50:44 # It is necessary for linking the library. Jul 01 11:50:47 jacques: Any special other libs I need for gphoto2? Jul 01 11:51:06 I didn't delete it, I swear! Jul 01 11:51:10 dunno. libtool docs just say libtool libraries are .la instead of .a Jul 01 11:51:11 ;) Jul 01 11:51:11 :) Jul 01 11:51:13 DaKa2, yes I have seen that. since I remove them and my linking still works, I think they are liars Jul 01 11:51:23 libtool are liars Jul 01 11:51:29 hehe Jul 01 11:51:33 there, I said it Jul 01 11:51:53 i have a hate-hate relationship with libtool Jul 01 11:52:05 "# libblah.la - a libtool library file" Jul 01 11:52:12 so it's a "library" in that it stores random stuff inside it Jul 01 11:52:41 NAiL, I don't recall needing anything other than libusb but I think there were optional components liek readline libexif expat Jul 01 11:52:41 a library, like books, text, you know? :-) Jul 01 11:52:55 Tiersten, so we are both right, they are libtool libraries and they are text files :-) Jul 01 11:53:05 jacques: yeah, jpeg, exif.. stuff like that Jul 01 11:53:05 yep Jul 01 11:55:34 configure: error: The pkg-config script could not be found or is too old. Make sure it Jul 01 11:55:44 pkg-config? Jul 01 11:55:56 ah yeah :-\ Jul 01 11:56:03 pkgconfig Jul 01 11:56:26 (is the package name) Jul 01 11:56:29 Isn't that built cross? I waguely remember that Jul 01 11:56:48 yep Jul 01 11:57:10 I think it's time to cook the ribeye Jul 01 11:57:27 That was a new expression ;) Jul 01 11:57:58 hmm Jul 01 11:59:52 "cook the ribeye", is that another way of saying "number 2"? Jul 01 11:59:52 :) Jul 01 11:59:58 or are you really cooking? Jul 01 12:00:03 no, it means cook the ribeye Jul 01 12:00:12 got any mustard? Jul 01 12:00:23 ~google ribeye Jul 01 12:00:24 get ribeye steak out of refrigerator, and cook it Jul 01 12:00:28 it's a cut of steak Jul 01 12:00:34 from the rib... Jul 01 12:00:40 take the rib out and you get the ribeye Jul 01 12:01:05 <[g2]> I thought that was a rib EEEEEE Jul 01 12:01:15 <[g2]> or rib OWWWWW Jul 01 12:01:45 yeah i know what a ribeye is, but it sounded like one of those expression of saying that you have to go to the toilet :) Jul 01 12:01:54 lol Jul 01 12:01:59 cook the ribeye? Jul 01 12:02:06 I don't want to know what you thought that meant exactly! Jul 01 12:02:11 LOL Jul 01 12:02:14 hehehe Jul 01 12:02:21 never MIND, puh-leaze Jul 01 12:02:41 I don't feel the need to mention here when I go to the toilet Jul 01 12:02:46 thought so Jul 01 12:02:52 hoped so Jul 01 12:02:53 tho I sometimes mention when I cook because it takes me AFK for a while Jul 01 12:03:04 man, even _I_ don't know what i was thinking Jul 01 12:03:08 can we drop this? Jul 01 12:03:13 urk Jul 01 12:03:17 :-) Jul 01 12:03:23 pkgconfig fails to build cross... ugh Jul 01 12:07:15 NAiL, hmm I installed mine from an ipk Jul 01 12:07:23 maybe it broke recently? Jul 01 12:07:29 maybe Jul 01 12:07:34 which version is it trying to build? Jul 01 12:07:42 0.15 Jul 01 12:07:57 ipkg status pkgconfig Jul 01 12:07:57 Package: pkgconfig Jul 01 12:07:57 Version: 0.15.0-r1 Jul 01 12:07:59 Quilt fails to patch, so it's probably something simple (Looks like wrong parameters to patch) Jul 01 12:08:04 -r1 Jul 01 12:08:04 ah Jul 01 12:08:16 lots of things breaking recently from the sounds of it Jul 01 12:08:19 Right now I'm too tired to look at it ;) Jul 01 12:08:27 It's way too hot here Jul 01 12:09:06 I feel a bit lazy, should probably look at sane Jul 01 12:09:22 dont really know where to start... Jul 01 12:09:56 What's wrong with sane? Jul 01 12:10:55 with working libusb its just what every other pixel is just a color Jul 01 12:11:01 in the scanned image Jul 01 12:11:12 I want to get gphoto2 working. It'd be neat to automatically empty my camera when it is plugged in :D Jul 01 12:11:15 and its b/w Jul 01 12:11:20 :) Jul 01 12:11:28 oh... neat bug Jul 01 12:12:02 yup, I'd love to troubleshoot that.... Jul 01 12:13:07 other than that it works with saned through the network Jul 01 12:13:32 It's libusbs fault :-P Jul 01 12:13:41 ... NO! Jul 01 12:14:04 i fixed that, atleast I think so.. Jul 01 12:16:35 from looking at the debian patch Jul 01 12:16:59 it appears it fixes the alignment issues, but doesn't address endian issues (if there are any) Jul 01 12:17:31 dammit.. I was never any good at endianess Jul 01 12:20:01 http://david.thg.se/saker/libusb/sane/tmp2.jpg <-- thats a scan of a 8cm CD Jul 01 12:20:09 notice its nice shape... Jul 01 12:20:28 nice inded Jul 01 12:20:32 indeed Jul 01 12:20:33 if it wasnt for the lines it would be round Jul 01 12:20:40 I think Jul 01 12:20:43 haha Jul 01 12:22:03 oh, looks like you scanned a PAL CD in NTSC Jul 01 12:22:15 lol Jul 01 12:25:50 well, a scan i grayscale mode is totaly perfect, so its somewhere in the color stuff Jul 01 12:28:57 is it possible that sane has alignment issues? Jul 01 12:29:07 because that's kinda what it looks like Jul 01 12:29:19 probably.. its just so big... Jul 01 12:29:30 dont know where to look Jul 01 12:30:17 Im just trying to find out if its a problem in the backend, in sane, or in saned.. Jul 01 12:32:17 same problem with scanimage on the slug, so not saned atleast Jul 01 12:33:53 optware build fails for me as follows (it was doing this right from the start - nothing has changed): Jul 01 12:33:54 + sh /home/nslu2/nslu2-linux/optware/toolchain/crosstool/crosstool.sh Jul 01 12:34:03 glibc refuses to build if LD_LIBRARY_PATH is set. Please unset it before running this script. Jul 01 12:35:48 Looks like optware needs some kind of 'setup-env' Jul 01 12:41:20 That jpg looks like someone passed 32bit data to something which was expecting 24bit. Jul 01 12:42:03 hm.. that could be it Jul 01 12:43:04 im digging in the internals of sane-backends, and i dont understand half of it Jul 01 12:44:07 I don't think SANE delivers JPEG itself (iirc - I only had to look at it once), so it's probably a mistake in whatever calls libjpeg Jul 01 12:44:30 actually, Im getting pnm from it, and converted to jpg later Jul 01 12:44:43 and that convertion was on a working x86 box Jul 01 12:45:49 Im starting to think its in the backend for my scanner Jul 01 12:46:03 I wish I had some other one, with another backend Jul 01 12:49:54 The aspect ratio is 4:3, the only obvious thing is if the NSLU2 backend produces 32bits whereas x86 gives 24 - perfectly possible. Jul 01 12:50:22 due to alignment Jul 01 12:50:26 constraints Jul 01 12:50:49 hm well... any idea what to look for? Jul 01 12:51:10 Im used to 100% x86 stuff, never had to bother with anything else Jul 01 12:51:13 Look at the PNM and see how many channels it has. Jul 01 12:51:43 Then look at the backend (and I don't know where) and see how many it produces. Jul 01 12:52:13 just add __attribute__ ((packed)) to every struct element (j/k) Jul 01 12:52:37 ;-) that doesn't do what you might hope on the ARM gcc Jul 01 12:53:10 I ended up doing something like that to get reiser4 working - it was a nightmare. Jul 01 12:55:35 hm.. the ppm subformat of pnm sane uses seems to be 24bit only Jul 01 12:56:41 That's correct - if it's PPM it's strictly three channels (not necessarily 8 bit per channel). Jul 01 12:57:05 Actually support for more than 3 is extremely limited... Jul 01 12:58:06 fun.. Jul 01 12:58:10 * DaKa2 understands 0 Jul 01 12:58:22 better google on arm porting Jul 01 12:59:14 I suspect it isn't ARM specific - it may just be broken whenever a back end produces >24bpp data. Jul 01 12:59:24 Just guessing... Jul 01 12:59:51 Anyone here know anything about openssl, the ipk for unslung doesn't appear to install the CA certificates Jul 01 13:00:00 probably, everybody uses x86 anyway, right? :) Jul 01 13:01:07 No, SANE was supposedly fairly portable, and x86 is normally the problem - because some of the original PC video cards were big-endian (on a little-endian CPU). Jul 01 13:01:38 (big-endian byte values, little endian-byte order). Jul 01 13:02:38 .. i hope they fired the ones who did that... Jul 01 13:03:46 Take a video chip developed for a 68000 and put it into a video card on an x86. Jul 01 13:04:19 Cheap, good engineering - the ARM has big-endian floating point values and a little endian CPU for the same reason. Jul 01 13:04:20 It wasn't necessarily that silly - Suns are big endian and I think SGI configured their mips processors as big endian so high end stuff needed to work on big endian architectures Jul 01 13:05:38 Right, where 'high end' means 'colour' - it doesn't matter at >4bpp Jul 01 13:06:08 (and that's why MACs are RGB while PCs are BGR ;-) Jul 01 13:30:32 mr_claus: ping? Jul 01 13:43:51 unslung-image now works (I.e. I just fixed the 'depmod-2.3' problem) Jul 01 13:44:49 NAiL: sysconf: if you use 'reflash' to flash the image (instead of RedBoot/upslug) all the configuration is saved. Jul 01 13:46:46 You're the MAN :D Jul 01 13:47:33 Hi, anyone here know if IPK_VERSION should be reset to 1 when the upstream package version changes Jul 01 13:48:02 AdamBaker, yes reset to 1 Jul 01 13:48:49 eno-away:thanks Jul 01 13:49:28 If I make a new version of a .bb (New sw release) of a package with MAINTAINER set, should I bug that maintainer, or set myself as maintainer? Jul 01 13:50:28 I've only dealt with "unmaintained" .bb's before, I think ;) Jul 01 13:54:44 Consult the maintainer I believe - after all the existing maintainer may be working on a new version too... Jul 01 13:54:57 ok :) Jul 01 13:55:03 kergoth: ping? ;) Jul 01 14:03:37 What I ended up doing with NTP is putting in 4.2.0 with settings to make it the default only for openslug. Guess that would work with maintained packages too if you don't get a response from the maintainer. Jul 01 14:04:15 kergoth is here once in a while, right? :) Jul 01 14:04:28 Right. Jul 01 14:04:35 I'm in no hurry. Jul 01 14:04:46 You could ask in #oe too. Jul 01 14:05:21 yeah. Gonna see if I can get it to build first though Jul 01 14:05:44 God aften :P Jul 01 14:05:59 #oe often works because some of these things are effectively centrally maintained - so koen or mickeyl may well respond. Jul 01 14:15:32 something else entirely... It seems anything I try to build now which uses quilt freaks out Jul 01 14:21:47 and I'm not getting anywhere with SANE, might have to give up for the moment Jul 01 14:25:48 [g2]: monotone is now stable - have you sent in your key? Jul 01 14:28:11 jbowler-zzz: did you push your depmod-2.3 fix? Jul 01 14:30:26 jbowler-zzz: I can "reflash" using glibc openslug to a uclibc openslug without breaking anything, right? Jul 01 14:39:54 <[g2]> rwhitby, I haven't sent my key in yet Jul 01 14:41:34 from raw master makefile build right now Jul 01 14:41:37 monotone: misuse: branch org.openembedded.nslu2-linux has multiple heads Jul 01 14:41:38 make: *** [openembedded/conf/machine/nslu2.conf] Error 1 Jul 01 14:42:47 rwhitby: fdb811870800bdc6cab69dcfe11f49ee8958d637 Jul 01 14:43:22 NAiL: yes, but you will want to do a 'turnup flash' before reboot or it probably won't come up Jul 01 14:43:44 (I.e. it does preserve the /linuxrc) Jul 01 14:44:40 aha Jul 01 14:45:44 dyoung-zzzz: mysterious, I just synced and I have only one head (the fd one above). Jul 01 14:47:02 jbowler-zzz: Yeah, I get the same head too Jul 01 14:47:07 I have the two Jul 01 14:47:23 merged and pushed Jul 01 14:47:48 and I got jbowler-zzz's change in the merge Jul 01 14:48:07 I someone has local commits - not synced - and they pull or sync they will grow a second head. Jul 01 14:48:14 (If someone) Jul 01 14:48:22 ok working now Jul 01 14:48:34 so what is the recommended way to push a change? Jul 01 14:48:47 vi file; mt commit, mt push, my sync ? Jul 01 14:48:49 <[g2]> anyone know what this means ? Jul 01 14:48:53 <[g2]> http://pastebin.ca/16671 Jul 01 14:49:24 <[g2]> I got it on the link of php5 -- I think something isn't configured properly for dynamic libararies Jul 01 14:49:26 dyoung-zzzz: you add your private key to the local database, then commit, pull, check heads, merge if necessary, then push. Jul 01 14:50:32 I'd recommend pull before commit, it will minimise the amount of branching. Jul 01 14:50:40 Ok Jul 01 14:51:10 Still, monotone clearly encourages the making of spaghetti Jul 01 14:51:25 Mmmm Spaghetti.... Jul 01 14:51:34 I Like Noodles. :-) Jul 01 14:51:57 jbowler-zzz: good point Jul 01 14:51:57 Spaghetti Creation Machine Jul 01 14:52:30 looks like we mght have a master build working soon .... Jul 01 14:53:00 The only problem I have at this moment is that LD_LIBRARY_PATH issue in optware. Jul 01 14:53:45 good thing that 2.4.11... is larger than 2.3r29 ... Jul 01 14:54:12 jbowler-zzz: we can add a setup-env script at the top-level in optware CVS, and then call that from the Master Makefile Jul 01 14:54:31 or just unset it in the optware Makefile - we do that with other vars in that Makefile Jul 01 14:55:13 2.4.22... I hadn't thought of that - it didn't matter for the build because I renamed the two .bb files. Jul 01 14:56:21 Ok, the two heads came from the propagate from org.openembedded.dev - until we propagate nslu2-linux back that will keep happening. Jul 01 14:57:19 jbowler-zzz: we should do that asap - are we ready for that? Jul 01 14:58:06 I think so, I've flashed my uclibc build - it seems fine. (Haven't tested unslung, optware-build hasn't completed, but optware is not really relevant here). Jul 01 14:58:41 (I haven't tested a true raw build either - I'm running 2.6.12 at present). Jul 01 14:59:45 bbl - must stock up on food Jul 01 15:00:03 rwhitby: Sorry. Haven't done it yet. I've been busy with work Jul 01 15:00:17 ok, once openslug-packages and unslung-packages build, I will propagate to oe.dev Jul 01 15:13:42 <[g2]> anyone run xxdiff ? I've got a 3-way merge and I don't know how do save :) Jul 01 15:14:36 was xxdiff what popped up on my end? Jul 01 15:14:46 kinda like vi, only three windows Jul 01 15:15:08 <[g2]> nod, but which file is the master the left ? Jul 01 15:15:13 xxdiff? Jul 01 15:15:19 <[g2]> yeah Jul 01 15:15:26 Hmmpf I dont get that, no GUI here Jul 01 15:15:40 cli Jul 01 15:16:43 as in, I got the cli version Jul 01 16:16:06 ok, org.openembedded.nslu2-linux has been propagated to org.openembedded.dev ... http://monotone.vanille.de/viewmtn/revision.psp?id=8e63001b71c44d7180fad5cb696c1f236ca499bf&ancestry_limit=0 Jul 01 16:20:09 viewmtn is pretty neat. Jul 01 16:20:23 the tree tracking graphic makes everything make sense. Jul 01 16:21:09 yeah Jul 01 16:22:05 as I said earlier, sorry, haven't had time to do it. busy with work stuff Jul 01 16:22:43 oh, that wasnt a stab at you tiersten. Jul 01 16:23:01 I just havnt quite seen a detailed tree, and think its pretty darn cool Jul 01 16:23:20 :) Jul 01 16:41:52 I think we are back to pre-bk-gone stability now ... Jul 01 16:42:07 and with a much better integrated build system Jul 01 16:42:32 back later Jul 01 17:35:37 [g2]: install meld Jul 01 17:36:14 Current versions of monotone (post 0.19) preferably used kdiff3 over either meld or the emacs/vile possibilities. Jul 01 17:36:26 <[g2]> meld ? Jul 01 17:36:51 It is possible to write a lua hook to handle the three way merge with the editor (or scripting) of your own choice... The actual code is in the source (i.e. the default hook). Jul 01 17:37:17 'meld' is a X based three way merge too which, despite being disliked somewhat on #monotone, is actually quite useable. Jul 01 17:37:20 <[g2]> I'm running 18 Jul 01 17:37:45 From this point of view it doesn't matter - 0.19 is the same. Jul 01 17:38:07 <[g2]> I only had a minor tweak to openslug-packages Jul 01 17:38:21 One important tip #monotone pointed out to me - if you get into trouble, kill the process - monotone is atomic, it won't damage the db. Jul 01 17:38:34 ewww gnome ;-) Jul 01 17:38:51 <[g2]> and the repo was several days old, so I just blew it all away and did a fresh wget etc.. Jul 01 17:38:58 jacques: yes, my thoughts too. You don Jul 01 17:39:06 't like it you write a lua hook ;-) Jul 01 17:39:55 <[g2]> jbowler-zzz, jacques What do you think about an OpenSlug release ? Jul 01 17:40:01 [g2]: it's going to happen again, to all of us. Unless you sync/update every second sooner or later a merge will be required. Jul 01 17:40:22 What for? Everything that has changed is in the feed. Jul 01 17:40:35 <[g2]> jbowler-zzz, Oh yeah the merge isn't a big deal Jul 01 17:40:49 [g2], probably a good idea, but personally I have a lot of catching up to do on the current monotone stuff Jul 01 17:41:18 <[g2]> me too on the monotone... what's that saying ? Jul 01 17:41:29 <[g2]> "Seek progress not perfection" ? Jul 01 17:41:33 damn gentoo Jul 01 17:41:39 Yeah, but what is the point of making a release - what needs to be reved that won't get reved automagically by the first ipkg upgrade? Jul 01 17:41:41 latest meld they have is from july 2004 Jul 01 17:42:01 jacques: that one works for me. Jul 01 17:42:13 but... but... it's so old Jul 01 17:42:13 <[g2]> jbowler-zzz, the point is I think we can make a pretty full featured stable release Jul 01 17:42:46 ? which is identical to OpenSlug-2.0??? Jul 01 17:42:47 <[g2]> I'm thinking OpenSlug, native compile, LAMP and a stable set of packages Jul 01 17:43:00 None of those are in the image. Jul 01 17:43:24 Tell me one thing which changes if a release is made now? Jul 01 17:43:37 <[g2]> actually, the script I tested with should be the same as in monotone now Jul 01 17:43:46 has kernel or glibc changed ? Jul 01 17:44:10 <[g2]> the switch to 2.3.5 with the glibc-linux headers automatically generated Jul 01 17:44:41 <[g2]> that was in OE a week or so ago and that's what I've been testing the native compile against Jul 01 17:45:14 mr_claus'es changes are in the kernel nslu2-io.c (udev support). Jul 01 17:45:28 I need to give the headers thing a hard look Jul 01 17:45:34 So the point of a release is to move to glibc 2.3.5? Jul 01 17:45:40 is it taking the kernel headers from our kernel ? Jul 01 17:45:58 <[g2]> it builds the 2.6.11 sanitized headers Jul 01 17:46:00 or is it doing some generic arm kernel config thing and using those headers? Jul 01 17:46:03 ah Jul 01 17:46:17 does perl pass all tests from those? Jul 01 17:46:47 <[g2]> nod Jul 01 17:46:58 <[g2]> all but 3 python Jul 01 17:47:01 <[g2]> and the kernel builds Jul 01 17:47:03 I guess they're OK then Jul 01 17:47:14 <[g2]> and apache and php5 Jul 01 17:47:32 <[g2]> and openssl Jul 01 17:47:39 what about kismet ? Jul 01 17:47:42 :-) Jul 01 17:47:46 <[g2]> haven't tried Jul 01 17:48:09 <[g2]> you guys know I'm having hw built right ? Jul 01 17:48:28 you might have mentioned that once or twice Jul 01 17:48:32 ;-) Jul 01 17:48:47 <[g2]> well there is a minipci slot Jul 01 17:49:05 <[g2]> and I'm sure there's tons of radio cards that'd work Jul 01 17:49:52 2.6.11 santized headers - I don't see the difference from before - what's changed? Jul 01 17:49:54 <[g2]> but kismet on WRT54G with the sd/mmc hack would probably be the ticked Jul 01 17:49:56 <[g2]> but kismet on WRT54G with the sd/mmc hack would probably be the ticket Jul 01 17:50:58 <[g2]> pb_ koen mickely or someone fixed the 2.3.5 glibc to automatically build them Jul 01 17:51:04 the headers in oe made configuring native dev easier Jul 01 17:51:09 <[g2]> or added the depends Jul 01 17:51:21 <[g2]> jacques, yes no more tarring Jul 01 17:51:26 yep Jul 01 17:51:56 <[g2]> I think we are at the point where a native development package would be pretty trivial Jul 01 17:52:06 The headers used in the build are the same as before. Jul 01 17:52:18 <[g2]> I didn't diff them Jul 01 17:52:26 <[g2]> but the appear to operate the same Jul 01 17:52:51 linux-libc-headers-2.6.11.1-r0 -> r1 Jul 01 17:53:16 I'm talking about the headers in /usr/include/[asm,asm-generic,linux] Jul 01 17:53:39 before those did not exist unless you manually tarred them up from your kernel build dir Jul 01 17:53:58 <[g2]> exactly and that was fixed in OE about a week ago Jul 01 17:54:07 <[g2]> I 've been testing against that Jul 01 17:54:32 [g2], I think you and I are on the same page on this, I was just answering jbowler-zzz Jul 01 17:54:45 <[g2]> Ahh.. ok Jul 01 17:54:46 <[g2]> thx Jul 01 17:54:55 No; I think I'm still not understanding the reason for a release. Jul 01 17:55:18 If linux-libc-headers is the issue, it can be added to the feed (now it makes a package - r0 didn't) Jul 01 17:56:22 It's not in openslug-packages. Jul 01 17:56:40 <[g2]> I doesn't need to be Jul 01 17:56:57 well before I can have a strong opinion either way I feel I need to learn how to checkout and build the latest code again Jul 01 17:57:07 True, it gets built anyway, but it needs to be in the dev package - and that doesn't exist yet. Jul 01 17:58:09 <[g2]> jbowler-zzz, I'm saying I think if we build what in monotone now and run the ipkg install script we probably have a full native devel environment Jul 01 17:58:10 So, as I understand it, the only reason for wanting glibc-2.3.5 was to get the libc headers, but in fact glibc-2.3.5 isn't required for that. Jul 01 17:58:53 <[g2]> I think that fix is only in 2.3.5 Jul 01 17:58:53 Well, linux-libc-headers-dev_2.6.11.1-r1_armeb.ipk should be in the feed. Jul 01 17:59:06 <[g2]> that's what get's installed Jul 01 17:59:30 It isn't in glibc, it's in linux-libc-headers. Jul 01 17:59:31 <[g2]> I think there's no a dependency on that Jul 01 17:59:57 Let me put it this way, linux-libc-headers-dev_2.6.11.1-r1_armeb.ipk should be in the feed now. Jul 01 18:01:22 And: RDEPENDS_linux-libc-headers: Jul 01 18:02:27 If it isn't in the feed then it is because the image modules are deliberately excluded, but I don't actually know how the feed gets built, jacques? Jul 01 18:03:45 not sure I understand the question - I thought feed was everything in openslug-packages Jul 01 18:03:53 I think it is too. Jul 01 18:04:12 as to how it's updated, I don't know, dyoung-zzzz might Jul 01 18:04:28 Here's the problem: if I do an ipkg upgrade now (well, when the latest stuff is in the feed) I suddenly get glibc in its entirety (I think). Jul 01 18:04:32 (how it's propagated to the server) Jul 01 18:04:43 ouch Jul 01 18:05:00 An interesting question at this point is who wants to be the openslug package manager... Jul 01 18:05:03 I don't recommend updating glibc with ipkg (at least it used to be a dicey proposition) Jul 01 18:05:25 new version of glibc, new kernel are decent reasons for a new release Jul 01 18:05:30 Right. So if glibc is about to get pushed to the feed, then we suddenly want to make a release... Jul 01 18:05:46 ok I understand your point now Jul 01 18:06:04 new kernel: I'm testing 2.6.12.2, I would like that to be in asap to avoid divergent kernel work. Jul 01 18:06:08 last thing we want is a lot of ppl doing ipkg upgrade and getting hosed because glibc is being replaced Jul 01 18:06:29 esp since sill ipkg won't tell you what it's about to upgrade Jul 01 18:07:23 Any guesses as to what the base module is called? Jul 01 18:07:40 libc6 ? checking Jul 01 18:08:04 Current build: libc6_2.3.5+cvs20050627-r0_armeb.ipk Jul 01 18:08:06 yeah I think it's libc6 Jul 01 18:10:00 OpenSlug-2.0: libc6_2.3.2+cvs20040726-r17_armeb.ipk Jul 01 18:10:42 Ok. OpenSlug-2.2 == kernel 2.6.12.2(or greater) + libc6_2.3.5 How does that sound? Jul 01 18:13:32 sounds good Jul 01 18:13:46 gotta go for a bit - BIAB Jul 01 18:20:46 <[g2]> sorry was away for a bit Jul 01 18:22:55 I'm running meld 0.9.4.1 Jul 01 18:23:23 <[g2]> jbowler-zzz, a big portion of my release thought is to have a known snapshot for users pulling from the ipkgs Jul 01 18:23:51 <[g2]> since much of that stuff doesn't need to be the daily flux Jul 01 18:24:21 <[g2]> then we can have a stable and controlled upgrade cycle for the vast majority of users Jul 01 18:24:28 <[g2]> maybe we are not there yet Jul 01 18:26:59 <[g2]> and OpenSlug package manager is needed Jul 01 18:31:32 The problem with kernel upgrades is I always forget to rebuild the eth modules - so 2.6.12.2 boots, but no ethernet... Jul 01 18:32:04 <[g2]> doh! Jul 01 18:33:53 To make it worse I now have to upslug. First time in several weeks. Jul 01 18:34:44 <[g2]> you've done great things for OpenSlug Jul 01 18:35:14 <[g2]> forgetting to build the eth modules isn't one of them :) Jul 01 18:37:05 Sometimes I wish bitbake had something like PVDEPENDS - i.e. rebuild me if the PV of this dependency changes. Jul 01 18:38:07 <[g2]> it's all a work in progress Jul 01 18:38:28 <[g2]> I'll tell you though it's night and day from a year ago Jul 01 18:38:59 <[g2]> I thought buildroot was cool then, but we've got a pretty slick setup now Jul 01 18:39:13 <[g2]> warts and all Jul 01 18:49:37 optware-build failure: (cd /home/nslu2/nslu2-linux/optware/builds; cvs -d:pserver:anonymous@cvs.sourceforge.net:/cvsroot/byrequest co -D "2005-04-23" byRequest;) Jul 01 18:49:37 cvs [checkout aborted]: end of file from server (consult above messages if any) Jul 01 18:50:19 There aren't any messages, but it happens every time (instantly). Jul 01 19:07:11 Kernel 2.6.12.2 is in monotone - not in the OpenSlug build yet, but can be built (see the Wiki). Jul 01 19:58:13 Hmm. 20 members of the slug special forces now. Jul 01 20:06:35 What does a package manager do anyway? Jul 01 20:08:24 He determines the suitablity of packages for openslug, then releases them to the package feed, and tasks relating to that. Jul 01 20:09:06 aha Jul 01 20:11:10 glc, theres a 2nd page too; unless you were counting de-underclocking technicians rather than raw number of de-underclocked slugs. Jul 01 20:12:22 Wow. The army grows. Jul 01 20:23:01 I added more power stats courtesy of yahpn to the wikipage Jul 01 20:23:24 I didnt make it pretty. If someone knows wikispeak, feel free to fix it. Jul 01 20:24:17 <[g2]> ~dyoung Jul 01 20:24:18 hmm... dyoung is just some guy with a UBESSA. That's a 'Unorthrodox But Effect Super Serial Adapter'. He is also the master of doing things in the ugliest possible way, and a part time kernel diddler. Jul 01 20:24:35 * dyoung-web takes a bow Jul 01 20:24:43 dyoung-web, your parts are on the way to me! Jul 01 20:24:44 <[g2]> ~praise dyoung Jul 01 20:24:44 All hail dyoung! Jul 01 20:24:59 My parts? Jul 01 20:25:06 Oh, my CPLDs and stuff? Jul 01 20:25:21 whatever it was that ka6sox got me to order Jul 01 20:27:59 were those your CPLDs then? Jul 01 20:28:12 sigh Jul 01 20:28:22 Probably. Jul 01 20:28:56 I miss a full day of nslu2-linux and feel completely out of sync.... what has happened in the meantime? Jul 01 20:29:03 How on earth am I going to change the uid/gid of a directory in an ipk? Jul 01 20:29:34 postinst Jul 01 20:29:41 aha! Jul 01 20:29:47 * NAiL hugs dyoung-web Jul 01 20:30:21 I gotta grab a beer, anyone else want one? Jul 01 20:30:32 yeah Jul 01 20:30:33 or Jul 01 20:30:42 * NAiL considers Jul 01 20:30:52 It's 5:30 am. A beer sounds tempting. Jul 01 20:31:16 well, you probably want to make sure the user/group exists before you change the ownership to them right? Jul 01 20:31:44 dyoung-web: yeah. And so far, the only way to add a user I know is to add it to base-passwd Jul 01 20:32:18 haven't quite gotten around to that "group" part yet. Guess it's somewhere ;-) Jul 01 20:32:48 I'm sure it will be well documented when get it all worked out. :-) Jul 01 20:33:04 jacques: ipkg -test upgrade will tell what it is going to do. -test is your friend Jul 01 20:33:05 hehe Jul 01 20:33:34 that isn't exactly common knowledge though. A lot of users are gonna stuff up on that I think. Jul 01 20:35:48 NAiL: yep, that's why we need an OpenSlug package manager. It's a harder job than Unslung, cause unslung doesn't change kernel or libc *ever*. Jul 01 20:37:44 * NAiL hides in the shadows Jul 01 20:41:49 Wouldn't it be a good idea to have a feed specifically for the openslug version in question? With only fixes backported? Jul 01 20:52:11 NAiL: that would be up to the OpenSlug package manager to decide :-) Jul 01 20:52:33 sounds like you've got some good ideas already ... Jul 01 20:52:42 heeeey Jul 01 20:52:48 don't look at me that way :-p Jul 01 20:53:05 I've heard this speech before! Jul 01 20:54:58 heh, back later. Jul 01 20:55:27 I'm not entirely sure I am up to the task. Jul 01 20:57:45 that's just great... my daily nlsu2-linux mailing message got cut at message 8 because *someone* couldn't figure out how to snip his own messages (or get his messages all in one posting) Jul 01 21:03:28 hey josh Jul 01 21:03:39 have you looked at the new master makefile yet? Jul 01 21:05:14 hi rwhitby, no i haven't had a chance Jul 01 21:06:23 ...looking at it now Jul 01 21:09:05 looks like it works ok to download and start building optware. i don't have monotone yet, so can't test anything else Jul 01 21:09:24 ...i'll be back in Davis soon and will be back in the swing of things Jul 01 21:09:53 It doesnt complete the optware part for me, croaking on fontconfig Jul 01 21:10:24 spews out all sorts of SGML warning/error messages Jul 01 21:11:03 hmm... interesting. likely a problem with fontconfig.mk itself rather than the master makefile i'd say Jul 01 21:11:14 can you quote a sample of the error messages? Jul 01 21:13:39 Yeah, just a moment. Jul 01 21:15:14 byRequest has a download problem too. Jul 01 21:15:39 nsgmls:/home/dereky/nslu2-world/optware/builds/fontconfig/fc-cache/../fc-cache/fc-cache.sgml:1:59:W: cannot generate system identifier for public text "-//OASIS//DTD DocBook V4.1//EN" Jul 01 21:15:40 nsgmls:/home/dereky/nslu2-world/optware/builds/fontconfig/fc-cache/../fc-cache/fc-cache.sgml:35:0:E: reference to entity "REFENTRY" for which no system identifier could be generated Jul 01 21:15:40 nsgmls:/home/dereky/nslu2-world/optware/builds/fontconfig/fc-cache/../fc-cache/fc-cache.sgml:1:0: entity was defined here Jul 01 21:15:53 thats the typical message for several hundred lines Jul 01 21:16:01 then it eventually says... Jul 01 21:16:22 nsgmls:/home/dereky/nslu2-world/optware/builds/fontconfig/fc-cache/../fc-cache/fc-cache.sgml:180:65:E: element "EMAIL" undefined Jul 01 21:16:22 Can't call method "value" on an undefined value at /usr/share/sgml/docbook/utils-0.6.14/helpers/docbook2man-spec.pl line 249, line 1. Jul 01 21:16:22 make[7]: *** [fc-cache.1] Error 8 Jul 01 21:16:40 Its using some host SGML stuff which may be the problem Jul 01 21:17:00 dyoung, that's weird, fontconfig shouldn't be trying to use docbook stuff when configured --disable-docs Jul 01 21:17:43 Yeah, I seem to be th eonly one that has this problem. Jul 01 21:20:30 think i've found the problem. line 412 of fontconfig's configure.in sets an automake conditional based on whether docbook2html was found, not whether docs are enabled Jul 01 21:21:21 jp30 thanks/. Jul 01 21:21:39 Its no rush though, since as far as I can tell, I'm the only one having the issue. Jul 01 21:21:46 dyoung does it work if you set the environment variable ac_cv_prog_HASDOCBOOK=no Jul 01 21:22:37 if so, it would be appropriate to set that in fontconfig.mk Jul 01 21:23:08 blew away builds/fontconfig to try again Jul 01 21:24:27 well, it went on to libgd; so I'm tempted to say yes. Jul 01 21:24:58 and on to libxml, so I'm thinking the answer is is. Jul 01 21:28:59 i've checked in the fix to fontconfig. dyoung, you might want to check that your docbook utils install is functional. in any case optware shouldn't be using it on some systems, and not on others Jul 01 21:29:58 I dont use it usually so dunno. Jul 01 21:30:23 heh my solution was gonna be install ubuntu. Jul 01 21:31:38 hey, change of subject, but i was impressed with the reports of the turboslug modification. i'll try that on one of my slugs when i get home Jul 01 21:32:38 the fingernail clip method is supposed to be the easiest way Jul 01 21:32:41 a lot of effort by a lot of people went into that find. Driven by the irritation of seeing "only" 133bogomips when other ixp platforms show the full 266 Jul 01 21:33:12 it's just too bad that dyoung had to go get a butt kicking when beewoolie decided to do the mod Jul 01 21:33:50 For the ones I modified, I used the solder lift method so the resistor is still connected on one side in case I need it later. Jul 01 21:34:07 "tombstoned" Jul 01 21:34:29 btw, how was the butt kicking, dyoung-web? Jul 01 21:34:44 it was nice. Jul 01 21:34:53 you sadist you Jul 01 21:35:05 or is that masochist? Jul 01 21:36:01 thank you guys for the great job. I turned one of my two slugs into turboslug today and it works great! I don't remember when was the last time i used a soldering iron Jul 01 21:36:44 eno, thank YOU for all of these great package contributions youve made. Jul 01 21:37:18 i hope i can get monotone done this long weekend Jul 01 21:37:29 me too. Jul 01 21:37:36 then I can have my monotone repo on a slug Jul 01 21:37:41 the same slug that runs apache Jul 01 21:38:15 the checked in version builds fine, but there's some runtime problem Jul 01 21:39:05 sqlite schema version mismatch or something like that, need to look into it Jul 01 21:39:13 is there a mod_python for unslung/openslug yet? Jul 01 21:39:28 there's one for unslung Jul 01 21:39:48 Hmm... Youre tipping me back to unslung.... Jul 01 21:40:11 dyoung-web, don't you have mulitple slugs? Jul 01 21:40:25 yeah, I have 3. Jul 01 21:40:31 or you should temp me to the openslug land Jul 01 21:40:36 None are in use right now. Jul 01 21:40:50 But I turboed one last night and put a shiny new USB stick on it Jul 01 21:41:04 I did have 3, but used one to bribe a graphics pro... Jul 01 21:41:10 i'm actually trying to test the new master makefile and got some problem Jul 01 21:41:23 I'm thinking I need to get back to having 3 Jul 01 21:41:28 eno, what sort of problem did you run into? Jul 01 21:41:33 NOTE: package cron-3.0pl1-r2: task do_package: started Jul 01 21:41:34 ERROR: function do_install failed Jul 01 21:41:34 ERROR: see log in /home/slug/openslug/tmp/work/cron-3.0pl1-r2/temp/log.do_install.25535 Jul 01 21:41:49 It's missing install.sh, probably? Jul 01 21:41:52 Mmm... when was the last time you tried this? Jul 01 21:42:02 (or rather, when did you get the Master Makefile last? Jul 01 21:42:03 ) Jul 01 21:42:39 i just did a wget, or do i need to wipe out and start from scratch? Jul 01 21:43:14 If you started with one before Jun30 western hemisphere time, you'll need to start fresh. Jul 01 21:43:27 but you can save your dowloads directory Jul 01 21:43:45 ok, let me try that, thx for the tip Jul 01 21:44:12 argh Jul 01 21:44:21 /etc/localtime is missing Jul 01 21:45:20 hey, eno, py-bluez is promoted and going in the feed now-ish Jul 01 21:47:08 great! i actually haven't tested it runtime, a friend requested and he will test it Jul 01 21:54:13 back later Jul 01 22:10:22 is there a netstat in some package on unslung? Jul 01 22:18:49 nevermind, found it in net-tools Jul 01 22:48:48 jbowler-zzz, cool, ipkg -test is a new one for me **** ENDING LOGGING AT Fri Jul 01 23:59:57 2005