**** BEGIN LOGGING AT Sat Jul 30 23:59:56 2005 Jul 31 02:59:02 can anyone load kernel-module-input w/o getting Jul 31 02:59:04 input: Unknown symbol hotplug_path Jul 31 02:59:04 input: Unknown symbol add_input_randomness Jul 31 02:59:06 ? Jul 31 03:19:43 Matta: will give it a try later Jul 31 03:21:17 k Jul 31 03:34:51 ha, I got my setup to make the kernel modules Jul 31 03:35:23 apparently, defconfig / .config requires the lines to be in the right place, I thought you could have them anywhere in the file Jul 31 03:44:16 Matta: so the unknown symbols have gone away, or are still unknown? Jul 31 03:47:02 nope, unknown, my compile results in the same problem Jul 31 03:47:14 but now I can debug Jul 31 03:47:24 ;) Jul 31 03:49:34 hotplug is compiled in the default kernel, right? Jul 31 03:52:15 hmm, apparently it's an upstream bug in the kernel Jul 31 03:52:51 03marceln * 10unslung/ (make/dovecot.mk sources/dovecot/configure.in.patch): Jul 31 03:52:51 FIX from Adam Baker for cross building dovecot. Jul 31 03:52:51 There are still some problems with kerberos libries on the host os. Jul 31 03:53:03 guess I'll have to make a temporary patch until they get it fixed upstream Jul 31 04:00:48 Matta: we can put a patch in openslug Jul 31 04:07:10 k, I'll test it here, when it works I'll post it here Jul 31 04:07:29 where should I put .patch files? Jul 31 04:07:53 and is it enough to edit nslu2-linux/packages/linux/nslu2-kernel_2.6.12.2.bb to make it work? Jul 31 04:08:59 the patch will probably become obsolete with the next upstream kernel release Jul 31 04:12:55 Matta: yep, nslu2-kernel_2.6.12.2.bb is the one to add the patch into the SRC_URI. Jul 31 04:13:58 and where should the actual .patch be located? Jul 31 04:14:09 openembedded/packages/linux/nslu2-kernel/2.6.12/ ? Jul 31 04:14:20 03rwhitby 07org.nslu2-linux.bitbake * r928367aa... 10/ (6 files in 5 dirs): Updated to bitbake 1.3.2 Jul 31 04:14:23 03rwhitby 07org.nslu2-linux.dev * r428b3f81... 10/Makefile: Improved the import-bitbake target Jul 31 04:20:56 source patched, compiling Jul 31 04:21:02 let's see what happens Jul 31 04:34:07 yess, it worked! =) Jul 31 04:35:39 http://people.realnode.com/~mnordstr/nslu2/missing-exports.patch Jul 31 04:49:27 gtg, but with that patch and the usbhid, input, evdev, usbserial and pl2303 modules loaded, everything I need seems to be working, even under uClibc... now all I have to do is write the application for it ;) Jul 31 04:49:31 l8r Jul 31 05:31:29 ACTION Jul 31 06:04:23 03ccsmart 07org.openembedded.nslu2-linux * r2692c20b... 10/packages/initscripts/ (3 files in 2 dirs): Migrate per package configuration files in drop in directory. Jul 31 06:05:54 [cc]smart: what's the detail of that change? Jul 31 06:08:45 <[cc]smart> it is intended to make volatile work as originally discussed, by providing a directory where packages can drop in their configs. Jul 31 06:08:55 <[cc]smart> the previous one was just an intermediate step. Jul 31 06:09:18 ah, cool. Jul 31 06:19:23 rwhitby: I suggest adding that patch to the official kernel if the input module is available Jul 31 06:22:49 Matta: so just add the patch to nslu2-kernel/2.6.12 and add it to nslu2-kernel_2.6.12.2.bb N2K_PATCHES ? Jul 31 06:24:51 Matta: is that what you did? Jul 31 06:41:55 yup Jul 31 06:52:26 Matta: done Jul 31 06:52:44 03rwhitby 07org.openembedded.nslu2-linux * racedaeff... 10/packages/linux/ (2 files in 2 dirs): Added the missing-exports patch from Matta to the OpenSlug kernel, to resolve the missing hotplug_path and add_input_randomness symbols. Jul 31 06:52:47 03rwhitby 07org.openembedded.nslu2-linux * rfaa2fa21... 10/ (18 files in 10 dirs): Jul 31 06:52:47 propagate from branch 'org.openembedded.dev' (head 824eb36b8f7f8e9953dd561a9f464e50042c85e2) Jul 31 06:52:47 to branch 'org.openembedded.nslu2-linux' (head 2692c20b64cebd9a996a3ec7d793938f235c2cd6) Jul 31 06:53:59 great Jul 31 06:54:35 all going well it should make it automatically to the feed within half an hour Jul 31 06:55:24 k Jul 31 07:06:54 03marceln * 10unslung/make/dovecot.mk: Jul 31 07:06:54 Cross compilation works (Thanks Adam Baker) Jul 31 07:06:54 Added dependecy to openssl Jul 31 07:32:59 03nail 07org.openembedded.nslu2-linux * r84ca6b48... 10/packages/ (3 files in 3 dirs): Added client programs for controlling the the mpd music player from the commandline, courtesy of Hannes Reich Jul 31 07:33:05 03nail 07org.openembedded.nslu2-linux * r9bfa38ef... 10/packages/at76c503a/at76c503a-modules_cvs.bb: Fixed compilation for 2.4 and 2.6 (.o vs .ko). Fixes OE bug #190 Jul 31 07:33:08 03nail 07org.openembedded.nslu2-linux * r17e420c5... 10/packages/portmap/portmap_5-9.bb: Added portmap-utils RDEPENDS so nfs works out of the box. Fixes slugbug #229 Jul 31 07:33:15 whatthe? Jul 31 07:33:24 * NAiL curses Jul 31 07:41:23 03nail 07org.openembedded.nslu2-linux * r17cd385f... 10/packages/ (3 files in 3 dirs): disapproval of revision 84ca6b48d0e2960ee88c5a4d6c227b9cc53e4d46 Jul 31 07:44:54 03repvik * r100 10/releases/OpenSlug-2.3-beta/openembedded/packages/perl/ (11 files in 2 dirs): Get a working perl into the stable feed Jul 31 07:50:07 hi, trying to build unslung5-5 from source, but the build failed, cause ipkg-cl isn't in the staging dir.... any hints, why it isn't? Jul 31 07:51:25 03repvik * r101 10/releases/OpenSlug-2.3-beta/openembedded/packages/portmap/portmap_5-9.bb: Updated portmap in stable feed Jul 31 07:53:13 hey macsat Jul 31 07:53:53 macsat, I was wondering whether you wanted to update some packages in the optware/wl500g feed to include some of the setup stuff from your tutorials. Jul 31 07:54:55 Sure....if I knew how ;-) Jul 31 07:56:00 mkdir -p /home/slug ; cd /home/slug ; make setup-optware ; make build-optware-wl500g Jul 31 07:56:21 oops - should be a wget http://www.nslu2-linux.org/Makefile in there as the third command Jul 31 07:56:54 you need to have the Asus toolchain (plus Oleg's extensions) installed in the usual place. Jul 31 07:58:39 e.g. the "Configuring the web server" section of http://www.macsat.com/php_web.php could just be included in the default thttpd.conf which is installed by the package. Jul 31 07:59:20 same with vsftpd Jul 31 08:00:08 and the rrdtool.sh could be included in the ipk too Jul 31 08:00:44 as well as making it depend on xinetd and making it drop in a inetd file in /opt/etc/xinet.d/ automatically. Jul 31 08:04:47 I'll try to take a look at it..... Jul 31 08:15:49 03nail 07org.openembedded.nslu2-linux * rf99b0f81... 10/ (4 files in 4 dirs): Back out RDEPENDS Jul 31 08:20:36 03nail 07org.openembedded.nslu2-linux * reab45770... 10/ (4 files in 4 dirs): disapproval of revision f99b0f8157e1cebc59aada3ef6513fcf8e724c48 Jul 31 08:25:41 03nail 07org.openembedded.nslu2-linux * r8fb3851e... 10/packages/portmap/portmap_5-9.bb: Correct commit. Back out the RDEPENDS from portmap Jul 31 08:29:18 <[cc]smart> is it possible to call the default do_install from within do_install ? if so, how ? Jul 31 08:30:26 [cc]smart: try do_install_prepend and do_install_append Jul 31 08:31:23 <[cc]smart> _prepend and _append are run during bitbake phase ... correct ? Jul 31 08:32:51 <[cc]smart> meaning not during ipkg install Jul 31 08:33:21 <[cc]smart> just as do_install is Jul 31 08:45:48 [cc]smart: base_do_install - look at the run.do_install* files in the working directory Jul 31 08:48:14 <[cc]smart> thx. meanwhile i tries _append and it seems to do what was needed. Jul 31 08:53:57 Someone made a comment on #oe to DaKa2 a few days ago about using/not using _append - I'm not sure what the issue was. Jul 31 08:54:38 I can't see any issues for do_install - maybe it was the pre/postrm stuff? Jul 31 09:23:24 <[cc]smart> NAiL: starce might be added to openslug-pacakges.conf. it works for me at least. Jul 31 09:23:32 <[cc]smart> s/starce/strace/ Jul 31 09:23:34 aah Jul 31 09:23:35 yeah Jul 31 09:23:48 Dunno why that was removed. I added it not too long ago. Jul 31 09:23:50 jbowler-away: ping? Jul 31 09:24:07 <[cc]smart> i say freeze is of hidden evil :P Jul 31 09:24:36 <[cc]smart> manually changing the distro files seems better to me, freeze is for individual developer needs :)$ Jul 31 09:28:37 NAiL: pong Jul 31 09:29:00 jbowler-away: Was it you who removed strace a while back? Or is my memory wrong? Jul 31 09:29:02 <[cc]smart> NAiL: timezones should also be added Jul 31 09:29:07 [cc]smart: why? Jul 31 09:29:21 NAiL: I also believed I had added it back... Jul 31 09:29:22 <[cc]smart> er, why not. sth. better ? Jul 31 09:29:53 [cc]smart: look in /etc/profile on recent (ie. since yesterday) builds Jul 31 09:30:16 NAiL: jacques asserted that it didn't work (as opposed to build) because of a missing patch - but that was a while back. Jul 31 09:30:47 Well, lets look at the log... I've been wondering about this too. Jul 31 09:31:07 How can I see the log for a specific file? I haven't figure that out yet ;-) Jul 31 09:31:30 The log is easy, it's the diffs I don't know about - monotone log packages/meta/openslug-packages.bb Jul 31 09:31:50 (log requires the path within the database, not a file system path) Jul 31 09:31:58 ah Jul 31 09:32:57 People keep saying 'monotone-viz', but last time I tried to build it (gentoo) it looked like I would be overrun by gnomes Jul 31 09:33:12 rofl Jul 31 09:33:39 Hum, it was me, on Date: 2005-07-01T07:39:09 Jul 31 09:33:55 care to re-add it? The last time I tried strace it worked perfectly Jul 31 09:34:00 <[cc]smart> TZ lacks the support zoneinfo gets. which is that zoneinfo is nicely managed by people out there to always be up to date with daylight saving time and all this, which can get quite complex... Jul 31 09:34:09 <[cc]smart> i think TZ is a nice default for embedded though Jul 31 09:34:13 Will do - I need to check it on uclibc too. Jul 31 09:34:18 <[cc]smart> so i would not replace it Jul 31 09:34:26 <[cc]smart> but certainly i'd offer zoneinfo Jul 31 09:34:33 <[cc]smart> as an option Jul 31 09:34:51 [cc]smart: fair enough. My TZ patch checks if /etc/localtime exists, so if you install zoneinfo you'll automagically get the right TZ info Jul 31 09:35:02 I haven't bothered trying to build it though Jul 31 09:35:12 <[cc]smart> BTW: can TZ also be set as default in /etc/localtime ? IIRC this should be possible. in this case it would be preferred over /etc/profile since it's there for all users, including system users. Jul 31 09:36:34 <[cc]smart> i always used zoneinfo before, so i'm quite sure it will work Jul 31 09:36:41 <[cc]smart> if not been fiddled with too much Jul 31 09:36:44 <[cc]smart> :) Jul 31 09:37:04 What do you mean by setting TZ in /etc/localtime? Jul 31 09:37:30 glibc reads /etc/localtime (which is usually a symlink to /usr/share/zoneinfo/somewhere) Jul 31 09:37:42 those files aren't exactly human readable IIRC Jul 31 09:37:46 <[cc]smart> IIRC it was possible to write the TZ string into /etc/locatime and it would get interpreted. or sth. along the lines of that Jul 31 09:38:16 <[cc]smart> zoneinfo files are "compiled" but you can disassemble them with a tool Jul 31 09:38:22 aha Jul 31 09:38:55 /etc/profile is used by all users Jul 31 09:39:31 <[cc]smart> in cases where daemons are started via /etc/init.d and then drop priviledges to different user without shell too ? Jul 31 09:39:33 [cc]smart: the nice thing about using the TZ string is that users can set the TZ value themselves Jul 31 09:39:34 <[cc]smart> i'm in doubt Jul 31 09:39:41 Hacking /etc/localtime into the firmware would prevent it being changed by zoneinfo Jul 31 09:39:54 <[cc]smart> NAiL: yes. the setting would just be the default Jul 31 09:39:55 [cc]smart: glibc reads the TZ var, then decides where to get the info Jul 31 09:40:12 the TZ var defaults to ":/etc/localtime" Jul 31 09:40:19 <[cc]smart> TZ contains the info Jul 31 09:40:39 if set to something else than :/etc/localtime Jul 31 09:41:44 <[cc]smart> now you set TZ in /etc/profile, afaik writing this string as content of 7etc/localtime does the same, just for all users. they stiil can override TZ with their .profile Jul 31 09:41:57 <[cc]smart> but it might also be a different file than /etc/locatime Jul 31 09:42:13 <[cc]smart> at least i'm pretty sure there is sth. along the lines of that Jul 31 09:42:20 <[cc]smart> as systemwide default Jul 31 09:42:37 I know the current solution works. It's not broken. Why fix it by creating another file? Jul 31 09:42:57 <[cc]smart> i think it's broken for those cases as descibed above, where daemons drop proiviledges Jul 31 09:43:30 <[cc]smart> they wont read profile Jul 31 09:43:35 <[cc]smart> afaik Jul 31 09:44:00 If you can show me that it doesn't work, I'll have a look at it. Jul 31 09:44:01 <[cc]smart> maybe daemons in general even Jul 31 09:44:55 but do feel free to get zoneinfo working. I haven't tried that, and some people might prefer that, since the TZ var is somewhat difficult to parse Jul 31 09:45:22 One thing I *do* know is that this solution works with *both* glibc and uclibc Jul 31 09:45:33 whereas the /etc/localtime one won't work with uclibc Jul 31 09:45:52 <[cc]smart> zoneinfo works. i've always used it before Jul 31 09:46:08 not on uclibc Jul 31 09:46:21 uclibc doesn't read /etc/localtime *at all* Jul 31 09:46:40 <[cc]smart> then it would get time for them to fix that Jul 31 09:46:46 no Jul 31 09:47:09 <[cc]smart> TZ is insufficient for production use Jul 31 09:47:10 because uclibc is supposed to be small, and there is support for setting the timezone info using the TZ variable Jul 31 09:47:25 why is TZ insufficient? Jul 31 09:47:47 <[cc]smart> because localtime is more detailed about daylight savings time Jul 31 09:47:57 how is it more detailed? Jul 31 09:48:02 <[cc]smart> it even does histroical calculation correct Jul 31 09:48:07 <[cc]smart> for example Jul 31 09:48:26 <[cc]smart> if go to zoneinfo and check the definitions, you'll see that theres many differences per country Jul 31 09:48:56 03sjmelhuish * 10unslung/make/oww.mk: Updated to source version 0.81.6 (adds hygrochron support and a bug fix) Jul 31 09:48:57 <[cc]smart> this simple picture of kind of vertical lines on the globe making timezones is far from the truth Jul 31 09:49:01 which is fair enough, but needed by a fairly small subset of users, which then should be running glibc and zoneinfo Jul 31 09:49:19 <[cc]smart> everybody who runs a server should care for correct time Jul 31 09:49:31 <[cc]smart> mail is such a thing Jul 31 09:49:40 <[cc]smart> you'd want correct time stamps Jul 31 09:49:40 Yes. And I do get correct time with the TZ variable Jul 31 09:50:11 Getting the correct time is not dependent on anything in the past Jul 31 09:50:15 <[cc]smart> in a few coutries mabe, in most not always Jul 31 09:50:24 <[cc]smart> wrong Jul 31 09:50:33 <[cc]smart> getting correct time is very dependetn on past Jul 31 09:50:43 [cc]smart: Give me an example of *any* country in which the TZ variable does not suffice. Jul 31 09:50:47 <[cc]smart> because the deifnition for time is changed from time to time Jul 31 09:51:01 <[cc]smart> NAiL: count pretty much ALL Jul 31 09:51:19 [cc]smart: Give me *one* example. Jul 31 09:51:25 <[cc]smart> where are you from ? Jul 31 09:51:41 <[cc]smart> i'll go and check yours Jul 31 09:51:51 Norway Jul 31 09:52:00 <[cc]smart> give me a couple minutes Jul 31 09:52:12 CET+1CEST+2 Jul 31 09:55:23 Anyway, if you create /etc/localtime, the TZ variable will not be set and thus your implementation will work. Jul 31 09:55:34 (If what you say really does work) Jul 31 09:57:49 But it's not portable to uclibc, and unless you intend spending the time to fix that (Which I doubt will be included in the lib anyways), using /etc/localtime is a non-solution Jul 31 09:58:23 using the TZ var is an OK solution. If you want it exactly correct, use glibc and zoneinfo Jul 31 10:00:52 how does TZ implement half-hours? Jul 31 10:01:36 not sure about that, but afaik it can do that as well Jul 31 10:01:55 <[cc]smart> i didn't find the database yet, which i'd prefer over disassembling this i don't remember how this is done ( but will if needbe). as a rpimer, this gives an idea where the issues are ( more to the end of the mail): http://lists.nyphp.org/pipermail/talk/2003-August/004657.html Jul 31 10:02:20 I find TZ confusing :) Jul 31 10:03:11 [cc]smart: yes? You can specify exactly when the change occurs. Jul 31 10:03:52 <[cc]smart> NAiL: and what you set is static Jul 31 10:03:58 yes Jul 31 10:04:12 the zoneinfo files are static as well. Jul 31 10:04:20 both require user intervention to update Jul 31 10:04:23 <[cc]smart> hwo do you tell TZ that before a given date DST was changed at 2am and after that date it was changed at 3 am Jul 31 10:04:31 you never know what time it is lithuania :) Jul 31 10:04:34 <[cc]smart> for example Jul 31 10:04:42 you don't. If you really need that, feel free to install zoneinfo Jul 31 10:04:52 <[cc]smart> that's what i'm saying Jul 31 10:04:57 <[cc]smart> i don't want the deafult changed Jul 31 10:05:02 I was there for two days before I realised that I was one hour ahead of the rest :) Jul 31 10:05:14 <[cc]smart> but would like zoneinfo added to openslug-pacakges.conf so that those who like it can use it Jul 31 10:05:19 of course Jul 31 10:05:28 if it compiles, I'll add it Jul 31 10:05:46 <[cc]smart> and suggested that when possible (for i thought i remebered that htis should be possible) TZ might be set in a system configfile instead of profile to ensure that everything has it Jul 31 10:06:00 <[cc]smart> it compiles for glibc Jul 31 10:06:09 <[cc]smart> don't know about uclibc Jul 31 10:06:22 doesn't matter if it compiles for uclibc. It won't work on uclibc. Jul 31 10:06:22 <[cc]smart> but does it make it worthless for users if it's glibc only ? Jul 31 10:06:38 uclibc doesn't support /etc/localtime at all. It *only* supports the TZ var Jul 31 10:06:48 hm, uh-oh.. are people moving over to uclibc? Jul 31 10:07:01 <[cc]smart> so you can zoneinfo make require glibc Jul 31 10:07:11 no, but some people run uclibc. And we try to maintain compatibility. Jul 31 10:07:21 [cc]smart: yes Jul 31 10:07:37 ah, ok Jul 31 10:07:40 or, put it in the "doesn't build when using uclibc" section of openslug-packages Jul 31 10:07:51 <[cc]smart> er, don't know abou thtis Jul 31 10:07:53 messy :) Jul 31 10:08:10 # These packages only build on TARGET_OS=linux, not Jul 31 10:08:10 # TARGET_OS=linux-uclibc Jul 31 10:08:16 <[cc]smart> actually didn't do anything with uclibc et, so didn't even know there might be an issue Jul 31 10:08:54 <[cc]smart> so can we do both these things ? add ztimezones to openslug-pacakges and the line you say, or would that not work out as expected ? Jul 31 10:09:07 is there a .bb? Jul 31 10:09:15 <[cc]smart> .bb ? Jul 31 10:09:24 for zoneinfo Jul 31 10:09:35 <[cc]smart> the package is called timezones Jul 31 10:14:06 strace seems to build fine on glibc - just testing on uclibc Jul 31 10:16:35 and uclibc Jul 31 10:16:37 pushing timezones change now Jul 31 10:16:49 <[cc]smart> NAiL: Norway seems to be an excellent example for changes in timedefinition Jul 31 10:16:56 NAiL: did you push groff earlier? Jul 31 10:17:41 jbowler-away: no, I think I held back on that, because it didn't build for me Jul 31 10:18:04 <[cc]smart> http://cvsweb.netbsd.org/bsdweb.cgi/src/share/zoneinfo/europe?rev=1.1.1.29.2.1&content-type=text/x-cvsweb-markup Jul 31 10:18:18 Ok, I see it in openslug-packages.conf but not in openslug-packages.bb - so it just got removed from my .conf Jul 31 10:19:11 jbowler-away: It was meant to go in with man and man-pages. But since I couldn't get it to compile I put it off Jul 31 10:19:19 I'm about to check in the ucslugc configuration - but that will happen in two steps because I will have merge issues (so I'll create two heads locally then merge). Jul 31 10:19:54 Yes, I had to shift 'man' to the not-uclibc section. Jul 31 10:20:20 ah, yes. I don't do uclibc builds at the moment Jul 31 10:20:29 Soon to change :) Jul 31 10:20:30 03nail 07org.openembedded.nslu2-linux * r6c6537eb... 10/packages/meta/openslug-packages.bb: Added timezones for glibc builds Jul 31 10:20:30 Let's see if groff builds today Jul 31 10:20:33 03nail 07org.openembedded.nslu2-linux * r121ddca6... 10/conf/distro/openslug-packages.conf: Forgot to add it here ;-) Jul 31 10:20:36 <[cc]smart> ii don't remember the syntax really, but there have been at least 4 changes for timedefinition in Norway this century Jul 31 10:21:01 <[cc]smart> er past :) Jul 31 10:21:13 [cc]smart: Now, if I really needed exact history of that I'd use glibc and zoneinfo. Jul 31 10:21:43 [cc]smart: but this is all historical, like I said, history may repeat but that history is gone. If you want to know the full horror look at www.astrodienst.com Jul 31 10:22:02 <[cc]smart> yes, as said, i wouldn't suggest to make this default. i just wanted to eyplain why i see a need for this Jul 31 10:22:38 Fair enough. timezones has been added, and will soon be in the feed Jul 31 10:23:08 <[cc]smart> i read a lot about this time stuff way back and am happy that only thing i need to know nowadays is that i should use zoneinfo Jul 31 10:23:44 <[cc]smart> thx Jul 31 10:23:55 If you read the link in the comments I added when the TZ variable is set, you will see that zoneinfo is necessary for historic use. But not all users require that. Jul 31 10:24:38 jbowler-away: groff still doesn't build on my end Jul 31 10:24:39 /home/repvik/unstable/glibc/openslug/tmp/staging/i686-linux/bin/install -c -m 644 img/pic* /home/repvik/unstable/glibc/openslug/tmp/work/groff-1.19.1-r0/image/usr/share/doc/groff/1.19.1/html/img Jul 31 10:24:43 /home/repvik/unstable/glibc/openslug/tmp/staging/i686-linux/bin/install-sh: img/pic* does not exist. Jul 31 10:24:46 make[2]: *** [install_html] Error 1 Jul 31 10:25:03 I haven't bothered to look at it at all Jul 31 10:25:14 Just saw that it was broken :P Jul 31 10:25:23 <[cc]smart> yes and i agree fully. but in your case, iirc you should add a defintion for the time of day when daylight savings time is switched to your TZ variable. Jul 31 10:25:41 [cc]smart: Well documented in that link. Jul 31 10:25:44 I can't say I'm greatly worried by groff or man - if I want to read man pages I use my local machine or the web. Jul 31 10:26:06 <[cc]smart> i'll use zoneinfo so i can ship around this Jul 31 10:26:10 jbowler-away: me too. bob_tm sent me lots of stuff that I added Jul 31 10:26:29 [cc]smart: feel free. It'll be in the feeds for everyone to use ;-) Jul 31 10:26:34 <[cc]smart> thx Jul 31 10:26:50 And for those that do not require the granularity, the TZ variable will automatically be set. Jul 31 10:27:03 (or the history) Jul 31 10:27:15 Ok, I'm good to go on the ucslugc config - this just changes OE, I will change the master makefile afterward (it wouldn't work the other way round!) Jul 31 10:29:15 on my wl-hdd /etc/TZ says EUT1EUTDST Jul 31 10:29:30 and the clock is 2 hours wrong, ofcourse :) Jul 31 10:29:32 heh Jul 31 10:29:38 <[cc]smart> /etc/TZ Jul 31 10:29:41 <[cc]smart> that's the one Jul 31 10:29:46 CET1CEST2 Jul 31 10:30:00 <[cc]smart> use this for default instead of profile should be a good option Jul 31 10:30:42 NAiL: and now it's 4 hours wrong :) Jul 31 10:30:47 hmm Jul 31 10:30:51 oh Jul 31 10:31:02 ah well.. better set it Jul 31 10:31:04 is that CET-1CEST-2? Jul 31 10:31:18 it now says CEST, so that's an improvement Jul 31 10:31:32 <[cc]smart> prolly yes, iirc TZ was exactly the other way round as everybody would usually think Jul 31 10:31:37 with CET-1CEST-2 I get correct time Jul 31 10:31:48 [cc]smart: just noticed that Jul 31 10:32:27 Don't know if the switchover date is correct though. But that's possible to specify as well. Jul 31 10:33:02 kolla: is wl-hdd glibc or uclibc? Jul 31 10:33:15 [cc]smart: setting /etc/TZ to echo CET-1CEST-4 >/etc/TZ Jul 31 10:33:17 argh Jul 31 10:34:02 setting that value on my glibc system in /etc/TZ doesn't affect anything. That does look like a uclibc thing to me Jul 31 10:34:49 <[cc]smart> both should be able to use it Jul 31 10:35:01 "should" is a keyword here Jul 31 10:35:08 <[cc]smart> but meanwhile i have the prrof, that setiing in /etc/profile is insufficient as i feared Jul 31 10:35:23 <[cc]smart> you can do the following to test: Jul 31 10:35:51 <[cc]smart> create an executable in /etc/init.d that does echo ${TZ} >/test.out Jul 31 10:35:53 [cc]smart: nope. setting that value in /etc/TZ doesn't do squat on my glibc slug Jul 31 10:36:26 <[cc]smart> well, on glibc i'd recommend zoneinfo :) but it really should work.... will try to identify why it might not Jul 31 10:36:53 This is a change to the *firmware*, there is no way zoneinfo should be in the firmware (even if it fits). Jul 31 10:36:56 <[cc]smart> back to the testcase, i now realise it might be insufficient as a proof Jul 31 10:37:23 <[cc]smart> no, as said zoneinfo should not be the default Jul 31 10:37:36 <[cc]smart> i fully agree on this Jul 31 10:37:46 Then what you are saying is irrelevant - NAiL just added the zoneinfo package. Jul 31 10:38:20 <[cc]smart> irrelevant ? if you say so, it will be. Jul 31 10:38:28 kolla: does 'date -u' output the correct (UTC) time? Jul 31 10:38:56 well, I have fixed it now, so yes Jul 31 10:39:08 just need to figure out how to set the hwclock Jul 31 10:39:24 you don't have hwclock? Jul 31 10:39:27 hwclock --systohc Jul 31 10:39:28 On openslug it happens automagically in /etc/init.d/hwclock.sh Jul 31 10:39:41 no, no hwclock here Jul 31 10:40:08 the anyways, no big deal Jul 31 10:40:28 Anyone know the work round for this: Jul 31 10:40:29 monotone: trying 3-way merge Jul 31 10:40:29 monotone: fatal: signal: SIGABRT (application abort requested) Jul 31 10:40:32 no working ntpclient either :) Jul 31 10:40:59 but that's ok, I set up a service on one of my boxes, so I now set time using date andbash Jul 31 10:41:12 jbowler-away: try using a different merge tool? ;) Jul 31 10:41:28 <[cc]smart> kolla: you have to easy options Jul 31 10:41:34 <[cc]smart> s/to/two/ Jul 31 10:41:41 date $(cat < /dev/tcp/smurf/1492) >/dev/null Jul 31 10:41:49 <[cc]smart> if i got you correct that you use ntpd on the lsug Jul 31 10:42:04 [cc]smart: kolla is talking about the wl-hdd Jul 31 10:42:09 yes Jul 31 10:42:29 on the slug I dont have any problems Jul 31 10:42:29 <[cc]smart> :) yes ntpd or yes wl-hdd Jul 31 10:42:31 kolla: it has samba, right? Do you have a somewhat synchronised samba server? :) Jul 31 10:42:34 <[cc]smart> ok Jul 31 10:42:48 samba? why would I use samba? :) Jul 31 10:42:53 to set the time :) Jul 31 10:43:00 overkill? :) Jul 31 10:43:30 I just output the format date wants on my smurf-box and let date on the wlhdd read it from there Jul 31 10:43:34 works fine Jul 31 10:44:05 I do th same on the zaurus, since its kernel (2.4.18) has this bug that makes ntpdate barf when IPv6 is active Jul 31 10:44:07 kolla: net time set ;-) Jul 31 10:44:45 command not found Jul 31 10:44:55 bleh Jul 31 10:44:56 I dont want samba.. *shrug* :) Jul 31 10:47:40 NAiL: meld worked, kdiff3 crashed, but the problem seems to be that your add of 'timezones' and mine of 'screen' et al in the same place causes problems. Jul 31 10:47:59 augh Jul 31 10:48:31 kdiff3 seems buggy :P Jul 31 10:49:07 it kept telling me that there was this one conflict that I hadnt fixed, and every time I fixed it and tried to continue, it again said that the conflict was there Jul 31 10:52:10 Yeah, well the commit I just synced has some duplicate packages in it - in both the glibc and the uclibc list - I'll fix that up too, but I need to commit now or I'll have the same merge conflict again but even worse. Jul 31 10:52:50 This may explain why strace apparently disappeared... Jul 31 10:53:09 (Depending on when it was added back in) Jul 31 10:53:24 More than a month ago I think Jul 31 10:53:38 Oh, ok, that was before I removed it. Jul 31 10:54:10 uh, no, can't be Jul 31 10:54:21 maybe two weeks or seo Jul 31 10:54:22 so Jul 31 10:54:32 Just remembered when I started using monotone :-P Jul 31 10:55:21 I removed it on July 1 (according to the log), so maybe this was a case of a bad merge. Jul 31 10:56:39 I was also sure I had moved 'man' to the glibc-only section too. Hum... Jul 31 11:00:43 The monotone server on nslu2-linux.org seems to be down Jul 31 11:01:33 apparently Jul 31 11:01:56 how is it restarted? Jul 31 11:02:41 it isn't running at all Jul 31 11:03:21 monotone: fd 8 (peer 66.159.209.54:44218) exchanged goodbyes and flushed output, disconnecting Jul 31 11:03:26 monotone: fatal: std::runtime_error: network error: recv failure: Connection timed out Jul 31 11:07:01 jbowler-away: it's back up Jul 31 11:08:19 I might have got the 'timezones' merge wrong Jul 31 11:08:34 I'm pulling now, let me see what my repo looks like Jul 31 11:08:42 funny, everytime I try to install the python ipk on the wlhdd, the entire thing hangs Jul 31 11:09:48 NAiL: I think you haven't got my changes because they were only on monotone.vanille.de (I hadn't been able to sync to nslu2-linux.org) Jul 31 11:09:56 ah, of course Jul 31 11:10:21 are you syncing them, or shall I? Jul 31 11:10:28 I started mtn on nudi again Jul 31 11:10:40 I've explicitly synced Jul 31 11:11:03 I still have a local change to openslug-packages.bb to remove the spurious duplication of man Jul 31 11:11:41 monotone: branch 'org.openembedded.nslu2-linux' is currently merged: Jul 31 11:11:41 9457b91acff6286ee7d3890e518852ab50658f9a jbowler@nslu2-linux.org 2005-07-31T17:46:29 Jul 31 11:11:57 That's a little while back, innit? Jul 31 11:12:28 That's the correct head (i.e. it matches mine) Jul 31 11:13:44 yeah, I see man is duplicated Jul 31 11:14:05 and there's no reason to have man-pages build on uclibc, so you might as well move that one Jul 31 11:14:33 03jbowler 07org.openembedded.nslu2-linux * rebd51a9e... 10/packages/ixp425-eth/ixp425-eth_1.1.bb: Remove the unnecessary and confusing copy of the patch Jul 31 11:14:38 03jbowler 07org.openembedded.nslu2-linux * r214e2822... 10/ (8 files in 2 dirs): Jul 31 11:14:38 Add support for UcSlugC builds. Jul 31 11:14:38 (1) the 'standard' settings in local.conf.template (from the nslu2-linux.org Jul 31 11:14:38 master makefile) have been folded back into the three distro .conf files. Jul 31 11:14:38 (2) openslug-image now builds for any NSLU2 distro Jul 31 11:14:38 (3) ucslugc configuration added Jul 31 11:14:40 (4) more comments in the openslug and ucslugc configuration files Jul 31 11:16:40 03repvik * r102 10/releases/OpenSlug-2.3-beta/openembedded/packages/portmap/portmap_5-9.bb: Back out that RDEPENDS since portmap is in the firmware image Jul 31 11:20:07 I just synced 174865e1a8d25de392e872a17add8d09431285ba to remove the duplications. Jul 31 11:20:31 03jbowler 07org.openembedded.nslu2-linux * r174865e1... 10/packages/meta/openslug-packages.bb: Remove duplicated packages. Jul 31 11:22:03 Do you know why monotone is twice, btw? I'm tempted to remove monotone-4 Jul 31 11:22:21 222222222222222222222222222222222222222222222000000000 0 Jul 31 11:22:33 indeed. NumPad? :) Jul 31 11:22:42 If you want to communicate with a pre-0.20 server you need to use monotone-4 Jul 31 11:22:57 ah Jul 31 11:23:03 Last time I checked 0.19 was still the default on Debian Jul 31 11:23:29 The suffix is the netsync protocol number. Jul 31 11:34:46 03nail 07org.openembedded.nslu2-linux * rc3ddcc36... 10/packages/meta/openslug-packages.bb: Moved manpages to glibc-only section and strace back to working Jul 31 11:40:51 03jbowler 07org.openembedded.nslu2-linux * rb6008583... 10/ (22 files in 9 dirs): Jul 31 11:40:51 propagate from branch 'org.openembedded.dev' (head ba788d2a3a02c778c4ebf930f757e20a54f2b642) Jul 31 11:40:51 to branch 'org.openembedded.nslu2-linux' (head 174865e1a8d25de392e872a17add8d09431285ba) Jul 31 11:41:12 <[cc]smart> hmmm... can't find the reason why cyrus-sasl does not work anymore. i would have thought it should be because of the change from db3 to db4, but i cannot pin it down.... sbdy maybe has a hint for me where to look into ? Jul 31 11:41:43 [cc]smart: no idea, atleast not until you are *somewhat* more specific as to how it fails. Jul 31 11:42:01 <[cc]smart> that's onw of the problems: it's called "generic failure" Jul 31 11:42:10 <[cc]smart> strace, at least to me also gives nothing usable Jul 31 11:43:09 what's ucslugc? Jul 31 11:43:16 uclibc slug Jul 31 11:43:26 ic Jul 31 11:43:44 uc Jul 31 11:44:01 so a slug with TARGET_OS_local = "linux-uclibc" ? Jul 31 11:44:05 yes Jul 31 11:44:13 nothing else? Jul 31 11:44:25 <[cc]smart> messages gives the most verbose note: saslpasswd2: unable to open Berkeley db /etc/sasldb2: Operation not supported Jul 31 11:44:27 yes - nothing else (all the openslug extras are not there) Jul 31 11:44:31 <[cc]smart> but it's still senseless to me Jul 31 11:44:43 k, so I'm running a ucslugc =) Jul 31 11:44:50 ah Jul 31 11:44:53 nice to know Jul 31 11:45:01 Matta: I disabled TARGET_OS_local - you have to use DISTRO="ucslugc" Jul 31 11:45:19 k, good to know Jul 31 11:46:50 03jbowler 07org.openembedded.nslu2-linux * rc462fc4d... 10/packages/meta/openslug-packages.bb: Merge the man/man-pages change Jul 31 11:52:18 hmm Jul 31 11:53:45 We both made the same change. Jul 31 11:53:49 yes Jul 31 11:53:52 which works fine Jul 31 11:54:52 Yes - I just put it on one line to keep the two packages together. Jul 31 11:55:51 ah, not exactly a bad idea that Jul 31 11:58:54 Success: root@lochalsh:~# df Jul 31 11:58:54 Filesystem 1k-blocks Used Available Use% Mounted on Jul 31 11:58:57 /dev/mtdblock4 6528 2984 3544 46% / Jul 31 12:00:07 Maybe I should remove netconsole.ko from the image too. It's 3748 bytes. Jul 31 12:00:19 er.. please dont :) Jul 31 12:00:27 or, well.. Jul 31 12:00:30 never mind Jul 31 12:00:32 I have disk Jul 31 12:01:06 To use ucslugc you have to add the root file system support back in. Jul 31 12:01:57 I won't remove netconsole though because it is part of openslug and it's actually in openslug-image - too much work to take it out for too little gain. Jul 31 12:10:16 03nail 07org.nslu2-linux.dev * r3d893ebd... 10/Makefile: Added setup-openslug-2.3-beta target Jul 31 12:15:31 03nail 07org.nslu2-linux.dev * r09536d7a... 10/Makefile: Added build-openslug-2.3-beta target Jul 31 12:15:34 03nail 07org.nslu2-linux.dev * ra66b4d4c... 10/Makefile: Added update-openslug-2.3-beta target Jul 31 12:16:19 * NAiL really starts to mess with the master makefile :P Jul 31 12:20:50 <[cc]smart> just noticed, that postfix depends to db3, so i went db3 for imapd and sasl too Jul 31 12:21:19 03nail 07org.nslu2-linux.dev * r3dc0db4f... 10/Makefile: Added upload-openslug-2.3-beta-cross target Jul 31 12:26:07 03nail 07org.nslu2-linux.dev * r567721ec... 10/Makefile: Added setup-openslug-2.3-beta-developer target Jul 31 12:53:28 hm, is the SECTION in the .bb-files used for anything? Jul 31 13:13:28 03nail 07org.nslu2-linux.dev * r89e37d51... 10/Makefile: Set up SVN_SSH, and make setup-openslug-2.3-beta target use shared downloads Jul 31 13:22:49 hm, so I made a new package in openembedded/packages/ and made a symlink to it from openslug/oe-symlinks/.. yet, when I try to bitbake it, I get "nothing provides mypackage" Jul 31 13:22:58 what am I missing? Jul 31 13:23:05 oe-symlinks aren't used anymore Jul 31 13:23:10 oh Jul 31 13:23:19 so.. I can just delete it? Jul 31 13:23:27 bb -b path/to/package/package.bb to build it Jul 31 13:23:31 dunno Jul 31 13:23:34 should be ok Jul 31 13:24:00 oe-symlinks is replaced by openembedded/conf/openslug-packages.conf Jul 31 13:25:47 (and bb freeze/unfreeze) Jul 31 13:26:01 What packages did you add? Jul 31 13:26:11 none yet Jul 31 13:26:36 think you missed a "distro" up there btw Jul 31 13:26:56 03nail 07org.nslu2-linux.dev * r059ac08d... 10/Makefile: Make "make update" update openslug-2.3-beta if it exists. Jul 31 13:27:08 yes, I missed distro ;) Jul 31 13:27:14 just checking to see if you're awake :P Jul 31 13:27:59 well.. Jul 31 13:30:24 why are such things changed all the time in the oe projects.. sigh Jul 31 13:31:04 In an effort to better the products :) Jul 31 13:31:42 By removing oe-symlinks, openslug can build entirely without berlios (which isn't exactly stable) Jul 31 13:33:00 well, IMO the entire OE community should come together and find out what they want - and stick with it :) Jul 31 13:33:25 Now that's a grand task Jul 31 13:33:48 and how we handle packages in openslug is of no concern for other OE developers Jul 31 13:34:06 (unless ofcourse we find a brilliant way to handle it) Jul 31 13:37:14 well, oe is so darn close gentoo portage, one could rather join the gentoo-embedded effort and leave ipkg behind :) Jul 31 13:37:48 but I know that wont happen anyways.. Jul 31 13:39:08 *shrug* I only get parse errors from my .bb :P Jul 31 13:39:35 03nail 07org.nslu2-linux.dev * rd480598f... 10/Makefile: Fix setup-openslug-2.3-beta-developer Jul 31 13:40:50 ah right Jul 31 13:40:53 a missing \ Jul 31 13:41:46 nah.. still not Jul 31 13:42:01 huh Jul 31 13:42:13 oh.. Jul 31 13:42:23 what was that about being awake again? :) Jul 31 13:42:47 heh Jul 31 13:42:51 apparently not awake enough Jul 31 13:44:41 03repvik * r103 10/releases/OpenSlug-2.3-beta/test: Just to test setup-openslug-2.3-beta-developer target Jul 31 13:46:14 03repvik * r104 10/releases/OpenSlug-2.3-beta/test: Remove it again ;-) Jul 31 13:47:33 Is something wrong with cvs as sf? Jul 31 13:47:42 I can't update a unslung package. Jul 31 13:52:30 ~emulate rwhitby Jul 31 13:52:30 Another Satisfied Customer! Jul 31 13:52:33 duh Jul 31 13:52:42 "My work here is done" Jul 31 13:52:56 Well. not exactly. But now I'm off to see a movie instead ;-) Jul 31 13:54:11 what movie? Jul 31 13:56:09 Stickmen Jul 31 14:11:37 mf.. what variable in OE represents the packages/blurb/blurb-123 directory in packages/blurb/blurb_123.bb? Jul 31 14:12:37 oh, none.. I get it Jul 31 14:35:55 marceln: the ipkg feed server (also at sf.net) seems to be slow / dead so I think sf.net has problems tonight Jul 31 14:37:59 There are already bug reports about the cvs server. Jul 31 14:38:37 So it's not working normal. Jul 31 14:39:17 I just finished with dovecot. Thanks for your help with configure. Jul 31 14:43:34 Purely selfish motive with configure - I haven't got my unslung build directory on the slug and it seemed like a better solution to fix configure than do a fresh checkout on the slug Jul 31 15:01:23 i compiled and installed dovecot two days ago Jul 31 15:01:29 seems to work ok Jul 31 15:02:16 mmm he left Jul 31 15:06:12 hm, what was the make/bitbake to update the package index again? Jul 31 15:07:16 ah, founf it Jul 31 15:07:19 ..d Jul 31 16:00:09 03daka 07org.openembedded.nslu2-linux * rb1872f89... 10/conf/distro/openslug-native.conf: Initial distro for compiling nativly for openslug on the nslu2. Needs some work, but basically works. Jul 31 16:25:24 03nail 07org.openembedded.nslu2-linux * r0f708a1a... 10/conf/distro/openslug-packages.conf: Added groff and udev Jul 31 16:34:41 03jbowler 07org.nslu2-linux.dev * r7f413ba8... 10/ (5 files in 4 dirs): Jul 31 16:34:41 Sufficient structure to make a buildable ucslugc (only ucslugc-safe tested Jul 31 16:34:41