**** BEGIN LOGGING AT Sun Dec 16 03:00:02 2007 Dec 16 05:07:55 03rwhitby * r7463 10optware/trunk/make/net-snmp.mk: net-snmp: Fixed log location Dec 16 06:31:20 03bzhou * r7464 10optware/trunk/ (make/net-snmp.mk sources/net-snmp/SNMP_FREE-gcc4.patch): net-snmp: fixed an invalid lvalue in macro, problem only shows up with gcc4 Dec 16 06:32:11 03bzhou * r7465 10optware/trunk/make/ntop.mk: ntop: fix cross-compile BADNESS Dec 16 06:45:45 03bzhou * r7466 10optware/trunk/make/p7zip.mk: p7zip: 4.55 -> 4.57 Dec 16 16:02:57 hmm. It would be REALLY cool to be able to run this on a slug http://backuppc.sourceforge.net/ Dec 16 16:03:34 none of the perl module requiremements seem particularly problematic. Dec 16 16:04:00 I should be working on a LOT of other things, but I guess I'll give it a shot Dec 16 16:04:47 because what's a sunday for if not to sit down and build distribution packages.... Dec 16 16:15:52 wow, looks like none of the dependency packages are out there at all. This should be interesting... Dec 16 16:16:30 but I suppose it also just became a project for when I have more time Dec 16 16:18:15 sounds like a nice project Dec 16 16:19:25 and a perfect fit, if I can get the dependencies ironed out and a reasonable default config, it would make slugos a LOT more acessible as an alternative firmware Dec 16 16:19:59 do you have the complete list of dependencies? Dec 16 16:20:01 "install slugos. Install this package. Wait... and voala, you have a web based backup solution far mroe robust than the default linksys stuff Dec 16 16:20:08 it's on the page Dec 16 16:20:13 just a few perl modules Dec 16 16:20:31 http://backuppc.sourceforge.net/faq/BackupPC.html#requirements Dec 16 16:20:33 not too bad Dec 16 16:20:34 of course you know what I'm thinking Dec 16 16:20:41 but not something I have time for this instant Dec 16 16:21:06 what are you thinking? Dec 16 16:21:13 optware, of course Dec 16 16:21:26 * hillct is a big fan of unstable over optware Dec 16 16:22:01 I don't like having things in divergent file structures Dec 16 16:22:13 * hillct checks the optware feed for kicks Dec 16 16:23:32 i can understand why people like slugos packages better than optware Dec 16 16:24:07 bitbake is just so... pretty Dec 16 16:24:11 :) Dec 16 16:24:37 and packaging cpan modules is almost automated Dec 16 16:24:43 so it shouldn't be a huge problem Dec 16 16:25:00 that's good to hear Dec 16 16:25:38 this is the sort of thing I'd dive into on a plane or (since I'd need connectivity) while kicking back over the holiday Dec 16 16:26:46 Optware seems to have Compress::Zlib but not the others Dec 16 16:26:54 so no big advantage there... Dec 16 16:27:15 it has archive-zip too Dec 16 16:27:22 oh Dec 16 16:28:02 yeah Dec 16 16:28:03 111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 Dec 16 16:28:09 oh soory :-( Dec 16 16:28:12 ??????????????????????????????????????????????????????????? Dec 16 16:28:23 I sais sorry m8 Dec 16 16:28:44 "all I want for christmas is a new keyboard" Dec 16 16:29:21 blame it on the stuck keys, encrusted with dirt Dec 16 17:23:37 eno: wow. the OpenEmbedded cpan class is very impressive Dec 16 17:24:32 I just need to get all the RDEPENDS ironed out which will be the trickiest part Dec 16 17:29:27 cool Dec 16 17:41:05 it's interesting. No error was generated when I built the perl-module-compress-zlib recipe without rdepends Dec 16 17:41:38 although presumably, the various modules this one uses, must be listed Dec 16 17:42:06 unless they're packaged with Perl and it appears they're all in separate ipks Dec 16 17:42:16 it should just be zlib and perl Dec 16 17:42:45 DEPENDS sure, but Compress::Zlib also uses IO::Compress, etc Dec 16 17:43:02 File::RsyncP seems a bit more tricky Dec 16 17:43:24 that's a real good test of OE cpan class Dec 16 17:43:42 yeah Dec 16 17:43:50 see Use lines http://search.cpan.org/src/PMQS/Compress-Zlib-2.008/lib/Compress/Zlib.pm Dec 16 17:44:05 lots of what are effectively runtime dependencies there Dec 16 17:45:16 good point Dec 16 17:46:43 though RsyncP doesn't have any unreasonable runtime dependencies Dec 16 17:47:11 possibly the only dependency there is rsync itself Dec 16 17:47:15 one of the subdir is autotool style configure Dec 16 17:47:27 oh, fun Dec 16 17:47:30 missed that Dec 16 17:55:02 actually, that's handled rather elegantly by inheriting autotools then subclass the configure and install methods Dec 16 17:57:03 the perl-module-authen-pam recipe demonstrated this approach Dec 16 18:05:04 oh joy Dec 16 18:05:37 03bzhou * r7467 10optware/trunk/make/perl-file-rsyncp.mk: added perl-file-rsyncp Dec 16 18:05:42 the flash crive on which I have my root filesystem isn't playing nice Dec 16 18:06:05 bbl, warriors game is on ... Dec 16 18:06:13 my slug wouldn't boot Dec 16 18:06:59 pulled the drive, plugged it into my laptop and it automounted without issue. Any recommendations on restoring the superblock? Dec 16 18:36:46 03bzhou * r7468 10optware/trunk/make/transmission.mk: transmission: 0.94+svn4003 -> 0.96 Dec 16 22:33:05 eno: it appears there's an issue with excessively long file lists when rsyncing - just that the slug runs out of memory. Guess I could add swap but do I really want it swapping to accomodate an rsync? probably not Dec 16 22:34:07 hillct, joshin is running into the same issue Dec 16 22:34:19 even with HUGE swap files. Dec 16 22:34:56 interesting Dec 16 22:35:21 I was completely lost until I came across this http://lists.samba.org/archive/rsync/2007-July/018106.html Dec 16 22:36:28 not much to be done except (in my case of backing up /home) break up the job a bit Dec 16 22:36:49 so I'm doing individual user directories separately, and it seems to work Dec 16 22:37:56 another option that seems to work is avoid the remote end processing at all by mounting up the slug via NFS, or whatever your preferred network filesystem may be Dec 16 22:38:09 kind of a lame workaround though Dec 16 22:38:30 yep it is. Dec 16 22:54:44 03bzhou * r7469 10optware/trunk/make/noip.mk: noip: 2.1.4 -> 2.1.7 Dec 16 22:58:48 hillct: the easiest way to package backuppc probably is to run configure.pl in postinst Dec 16 23:00:05 I haven't dug into it since earlier this afternoon. How much configuration is done there? Dec 16 23:04:30 hmm. It would make packaging trivial but it would make uninstallation a real bitch Dec 16 23:10:47 eno: I'm definately going to try to avoid that approach, but yes, it would be easier. the OE guys gave me all manner of hell last time I suggested running a substantial installation script as part of a package postinstall **** ENDING LOGGING AT Mon Dec 17 02:59:57 2007