**** BEGIN LOGGING AT Thu Jun 22 02:59:56 2006 Jun 22 04:11:22 rwhitby: cpio keeps being cleaned up even though the package is done and dusted Jun 22 04:11:28 any ideas? Jun 22 04:12:50 that's usually either a version problem, or a makefile dependency problem Jun 22 06:11:20 03oleo * r3486 10optware/trunk/make/cpio.mk: cpio: proper ipk cleanup Jun 22 06:14:03 oleo: cool. I've been updating make files that seem to be based on an older template but I missed that issue. Jun 22 06:15:24 there are other make files that use the older approach to cleanup. I might tidyup some of them Jun 22 06:16:20 please do. any you can change from static control files to generated control files would be good to. Jun 22 06:16:35 yeah been doing that Jun 22 06:16:39 I would love to have only directories under the source directory and move any files there into subdirs. Jun 22 06:17:32 ah hadn't noticed the files in the sources directory. will fix some of those now Jun 22 06:17:54 need some easy wins after some recent optware updates! Jun 22 06:18:02 they're the really old ones ... Jun 22 06:23:07 rwhitby, what would you suggest for two different toolchains for two Oleg 0.9.19 and 0.9.27 uClibc based distros? A different target? Like wl500g and wl500gw? Later for 0.9.27 wchar firmware? Jun 22 06:23:54 Hmm - good question Jun 22 06:24:25 maybe we need to start versioning optware Jun 22 06:24:48 (the feed dirs that is( Jun 22 06:26:11 For wl500g this would be nice. But is this the case for other arch? Jun 22 06:26:27 not yet, but could be in the future at any time Jun 22 06:26:39 (we would just symlink those that didn't change) Jun 22 06:27:32 So the current stable feed would be 1.0, and the unstable feed would be 2.0, and the wl500g dirs would be different. Jun 22 06:44:56 What about versioning targets only? Jun 22 06:53:18 how so? Jun 22 06:58:43 I think like subarch markup. like OPTWARE_TARGET=wl500g and TARGET_FIRMWARE=1.9.2.7-7 Jun 22 07:07:41 I am negotiating with Oleg what should future firware based on uClibc 0.9.27 include. We agreed on wchar support for 1.9.2.7-8 but we disagree on locale support. Jun 22 07:08:19 so go with wchar for now and negotiate locale for a later update Jun 22 07:11:34 03placid * r3487 10optware/trunk/ (make/cvs.mk sources/cvs.control): cvs: rewrite make file to remove manual control file Jun 22 07:12:00 We have 1MB free space in firmware which he spends on tcpdump. Locale support brings 100k which means tcpdump should be removed and he don't wants to remove if. Negotiating for later in not an option from my point of view as this may mean that we would have two different /lib/uClibc and /opt/lib/uClibc libraries! Jun 22 07:13:58 bit tricky since it is a network router and tcpdump is probably pretty relevant Jun 22 07:14:50 is there a lightweight replacement for tcpdump or some ability to reduce size through removing features? Jun 22 07:19:32 I've had a look into tcpdump bud did not find any way to shrink it. I've said that it could be installed later with optware or fetched somehow. No success in persuaiding. I think that, if someone is so dumb not to be able simple network config, than even tcpdump will not help him. Jun 22 07:22:44 why do we need locale? Jun 22 07:23:05 (i.e. is it required to actually compile stuff, or just a nice to have for language support?) Jun 22 07:25:19 I am no expert in locale, but I see that if I include this in uClibc, many software compile out of a box. Otherwise one must findle with gnu libintl and friends. And I saw this is not an easy job. Jun 22 07:26:34 you've made this argument to Oleg? Jun 22 07:28:56 Yes. But Oleg suggested two libraries. Firmware /lib/uClibc w/o locale and Optware /opt/lib/uClibc with locale. And I see this as a deficiency. Jun 22 07:30:12 I think that one single /lib/uClibc is much better solution. Jun 22 07:31:37 03placid * r3488 10optware/trunk/ (make/bison.mk sources/bison.control): bison: rewrite make file to remove manual control file Jun 22 07:38:14 I suspect troubles with different rlink search paths if using two different uClibc. If using Oleg suggestion, we don't need firmware to upgrade to later version. We could move to 0.9.27 with all features in Optware and firmware could remain 0.9.19. Jun 22 07:39:27 rwhitby , what do you think of tis in terms of memory usage and search paths? Jun 22 07:49:56 memory is tight. you don't want to have two libraries in ram Jun 22 07:50:26 search paths won't be a problem, cause the already link in libs from /opt/lib in optware using the linker rpath flags Jun 22 07:54:10 How do I get an info of rpath in executable, to verify this issue? Jun 22 07:55:33 dunno Jun 22 07:55:47 ldd? Jun 22 07:55:56 objdump somehow Jun 22 08:07:51 one example which shows that RPATH is not in all executables. mipsel-uclibc-objdump -x zip |grep RPATH return nothing while mipsel-uclibc-objdump -x digitemp_DS2490 |grep RPATH Jun 22 08:08:05 returns RPATH /opt/lib Jun 22 08:14:00 right - the .mk file has to be correct to set it Jun 22 08:14:21 (and it needs to get though to the final link step in the source package makefile) Jun 22 08:31:57 Gone wild search for missing rpaths. The following packages misses RPATH: adduser, alac, antinat, audiofile, autoconf, automake, cogito, getmail, ghostscript, iozone, ipython, mrtg, netio, psutils, unzip, xmail, zip Jun 22 09:55:12 03placid * r3489 10optware/trunk/make/flex.mk: flex: upstream to 2.5.33 and update make template Jun 22 10:53:21 03placid * r3490 10optware/trunk/make/cpio.mk: cpio: increment package number to try and clear up autobuild issue Jun 22 11:25:59 Anyone knows how to fix the clock in the slug? (Running DebianSlug 3.10) Jun 22 14:29:30 hey did you guys have to write drivers for the NPE's on the slug? Jun 22 15:54:04 hey did you guys have to write drivers for the NPE's on the slug? Jun 22 16:01:06 hello? Jun 22 16:02:57 yeye Jun 22 16:03:01 dunno Jun 22 16:03:56 man, tryin to get those NPE's up is a nightmare. Jun 22 16:04:03 are any of the dev;s in the room? Jun 22 16:05:12 yup Jun 22 16:05:23 but i think they're not "present" atm Jun 22 16:06:22 hrmpf... that fucking samba ;/ Jun 22 16:07:42 nuts Jun 22 16:08:00 what trouble you havin w/ samba? Jun 22 16:08:31 getting started with it ;) Jun 22 16:09:07 what are you tryin to do? Jun 22 16:09:26 the complete thing i'm trying or just that samba thing? Jun 22 16:10:36 i'm trying to boot the slug from a usb stick... Jun 22 16:11:00 and when i plugin my usb harddrive i want a samba server which shares the hdd... Jun 22 16:12:13 got problems with adding users at first and such stuff Jun 22 16:12:43 ah Jun 22 16:14:27 i have to add normal users before i add samba users? Jun 22 16:15:57 i don't think so. Jun 22 16:16:05 have you looked at the samba handbook? Jun 22 16:16:31 i don't really use samba that much, only when i have to mount a winblows share Jun 22 16:16:43 otherwise i just use the Unix Services for Windows Jun 22 16:16:47 for the windows machines Jun 22 16:16:56 tried with the docs... failed... Jun 22 16:17:07 i think i'll try again this evening.. now: simpons ^^ Jun 22 16:17:12 simpsons i mean Jun 22 16:17:43 man, i haven't watched that in ages. Jun 22 16:19:06 hld: did you try smbpasswd to add the user? Jun 22 16:57:20 vmaster, yup Jun 22 16:58:41 root@NSLU2:~# smbpasswd -a andi Jun 22 16:58:41 New SMB password: Jun 22 16:58:41 Retype new SMB password: Jun 22 16:58:41 Failed to initialise SAM_ACCOUNT for user andi. Does this user exist in the UNIX password database ? Jun 22 16:58:41 Failed to modify password entry for user andi Jun 22 16:58:41 root@NSLU2:~# Jun 22 17:54:10 hld: add them as a linux user first Jun 22 17:54:21 then add them as a smbuser Jun 22 18:16:01 with adduser? Jun 22 18:17:43 yeah Jun 22 18:22:42 root@NSLU2:~# adduser andi Jun 22 18:22:42 adduser: group name andi already in use Jun 22 18:24:28 is there already a group "andi" define in /etc/group? Jun 22 18:31:09 hey did you guys have to write drivers for the NPE's on the slug? Jun 22 18:38:15 andi:x:500: Jun 22 18:38:23 from cat /etc/group Jun 22 18:39:53 okay, is there also a andi in /etc/passwd? Jun 22 18:49:46 nope Jun 22 18:52:29 any devs in here? Jun 22 20:22:23 any devs in here? Jun 22 20:22:41 where can i get a hold of your patches to the source tree Jun 22 20:23:13 kernel sources are in the monotone repository Jun 22 20:25:28 see http://www.nslu2-linux.org/wiki/Development/HomePage Jun 22 20:25:28 what is the monotone repository? Jun 22 20:26:00 i just want the source tree, none of the other stuff, or patches Jun 22 20:26:25 i am not developing around the nslu, i just need patches to get the npe's to work Jun 22 20:30:01 again, i just want the SOURCE TREE Jun 22 20:30:04 not all your other stuff Jun 22 20:30:15 i don't care about the packages and shit Jun 22 20:35:38 If you don't want the very latest version you can get the latest source code release from http://sources.nslu2-linux.org/releases/ Jun 22 20:59:11 thanks Jun 22 22:06:23 AdamBaker: actually, the latest patches are in SVN now Jun 22 22:13:07 Nail: I thought that was only for in progress rather than completed patches Jun 22 22:14:01 no, actually the kernel repo in svn is the one in use now. The patches are pulled from there during the building Jun 22 22:25:58 Do you know if there is any intention to put viewcvs on the svn server? Jun 22 22:27:20 uh? You can browse the source and commit history on trac.nslu2-linux.org Jun 22 22:29:08 Ah - I hadn't spotted that - thanks Jun 22 22:30:18 np Jun 22 23:12:09 03placid * r3491 10optware/trunk/make/cpio.mk: cpio: update make file based on more recent make template to fix clean problem Jun 23 01:27:18 03placid * r3492 10optware/trunk/make/cpio.mk: cpio: fix path issue in .built section Jun 23 02:35:19 03placid * r3493 10optware/trunk/make/cpio.mk: cpio: remove a typo from .built section **** ENDING LOGGING AT Fri Jun 23 02:59:57 2006