**** BEGIN LOGGING AT Mon Oct 15 03:00:02 2012
Oct 15 10:01:43 <phaidros> wiki will be down for one minute
Oct 15 10:04:07 <jow_laptop> ok
Oct 15 10:50:47 <phaidros> wiki back
Oct 15 12:45:18 <jwendell> hi, folks. I'm wondering what's the best way to update packages in openwrt. I mean, it's not just enough to put them in a repository, it's necessary to deal with changes in config file...
Oct 15 12:45:48 <jwendell> I saw that opkg saves the new config file with another filename if they differ...
Oct 15 12:47:00 <KanjiMonster> jwendell: it does so only if the config file differs from the (old) package provided one; if it's the same it just replaces it (afaik)
Oct 15 12:47:14 <jwendell> yep
Oct 15 12:47:34 <jwendell> in my case I don't want to end up with 2 config files
Oct 15 12:47:41 <jwendell> I'd like to merge them
Oct 15 12:47:47 <jwendell> (even if manually)
Oct 15 12:48:17 <jwendell> I thought about using uci-defaults to set some config, but it only runs on reboot
Oct 15 12:48:29 <KanjiMonster> you can add a post-install hook to the package
Oct 15 12:48:38 <jwendell> so, I'm thinking about put a lot of stuff in postinstall
Oct 15 12:48:59 <jwendell> I'm wondering if there's something more 'elegant' :)
Oct 15 12:50:43 <jwendell> is there any variable that expands into opkg's destdir?
Oct 15 12:56:09 <jwendell> I mean, if in postinst script I do 'cp /etc/....' it's technically wrong, once DESTDIR can be different than /
Oct 15 12:57:23 <jow_laptop> $IPKG_INSTROOT
Oct 15 12:58:41 <jow_laptop> oops
Oct 15 12:58:46 <jow_laptop> $PKG_ROOT is set by opkg
Oct 15 12:58:53 <jwendell> :)
Oct 15 12:59:05 <jwendell> thanks!
Oct 15 12:59:15 <jow_laptop> anyhow, you can answer such questions yourself by reading the opkg source
Oct 15 12:59:28 <jow_laptop> I too had to run  grep -r setenv build_dir/linux-*/opkg-*/
Oct 15 12:59:41 <jwendell> so, every script in postinst hooks should use that variable, right?
Oct 15 12:59:51 <jow_laptop> its futile
Oct 15 13:00:06 <jow_laptop> we do not really support alternative destinations
Oct 15 13:00:29 <jwendell> ah
Oct 15 13:00:40 <jow_laptop> it cannot work if the package is not prepared for it
Oct 15 13:00:50 <jow_laptop> it will still search itsl ibraries in /usr/lib
Oct 15 13:00:55 <jow_laptop> still search its configs in /etc
Oct 15 13:01:02 <jow_laptop> still place its pid in /var/run
Oct 15 13:01:03 <jow_laptop> etc.
Oct 15 13:01:10 <jwendell> I see
Oct 15 13:01:51 <jow_laptop> opkg --dest is mainly useful for preparing stuff liek chroots
Oct 15 13:02:01 <jow_laptop> or to prepare another rootfs
Oct 15 13:03:43 <jwendell> we could drop 'dest ram /tmp' from opkg.conf then
Oct 15 13:03:54 <jow_laptop> thats actually useful
Oct 15 13:04:41 <jow_laptop> I commonly use  opkg install -d ram install tcpdump; LD_LIBRARY_PATH=/tmp/usr/lib:/usr/lib:/lib /tmp/usr/sbin/tcpdump
Oct 15 13:04:49 <jow_laptop> for units with little ram
Oct 15 13:04:54 <jow_laptop> erm little flash
Oct 15 14:11:11 <KanjiMonster> danitool: I can't get your hg553 patch to apply, it seems to be seriously borked; can you resend it as an attachment (unzipped)?
Oct 15 14:11:20 <KanjiMonster> ah
Oct 15 14:11:22 <KanjiMonster> too slow
Oct 15 15:40:32 <russell--> my wgt634u build of r33760 went okay last night, so my net45xx kernel build must be something specific to it
Oct 15 19:36:33 <jwendell> hey guys, how are your relationship to opkg team?
Oct 15 19:36:56 <jwendell> I see there are local patches that are perfectly applicable upstream but are stick in our sources
Oct 15 19:37:04 <jwendell> is there a reason?
Oct 15 19:37:16 <jwendell> I mean, are they unresponsive, or dead?
Oct 15 19:38:05 <jwendell> example: 014-errors-to-stderr.patch
Oct 15 19:50:03 <jow_laptop> this patch was rejected upstream
Oct 15 19:50:31 <jow_laptop> I have commit access to the upstream repo but since there was no agreement on this patch I just kept it in openwrt
Oct 15 19:51:34 <jwendell> good
Oct 15 19:52:17 <jwendell> jow_laptop, why upgrade was restricted to a package?
Oct 15 19:52:36 <jwendell> (009-remove-upgrade-all.patch)
Oct 15 19:52:37 <jow_laptop> because it does not work as expected on most deployments
Oct 15 19:53:28 <jwendell> it's no the same as get the list-upgradable and call upgrade on them?
Oct 15 19:53:36 <jwendell> it's *not* the same as get the list-upgradable and call upgrade on them?
Oct 15 19:54:06 <jow_laptop> it is
Oct 15 19:54:27 <jwendell> so, why doesn't it work as expected?
Oct 15 19:54:37 <jow_laptop> because it fills up the overlay
Oct 15 19:54:42 <jow_laptop> it implicitely upgrades kmods
Oct 15 19:54:53 <jow_laptop> it gives users the impression that they can just dist-upgrade their firmware
Oct 15 19:55:04 <jow_laptop> which fails spectaculary and results in a brick 90% of the time
Oct 15 19:55:56 <jwendell> I see... I'm working on a jffs2-only firmware, which doesn't have this space-problem
Oct 15 19:56:39 <jwendell> and we are going to use the repository thing to update individual packages
Oct 15 19:57:09 <jwendell> (of course we are going to use our own repository)
Oct 15 20:01:27 <jow_laptop> ok
Oct 15 20:37:31 <written_tabletpc> what is a good supported router these days for openwrt?
Oct 15 20:45:15 <vanDrunen> several AR71xx routers (TPlink and D-link) and the Alix boards are excellent
Oct 16 01:41:14 * russell-- has built for wgt634u, alix2, ar71xx, all okay, x86-generic-net45xx not okay... some kind of kernel config problem
**** ENDING LOGGING AT Tue Oct 16 03:00:01 2012