**** BEGIN LOGGING AT Sat Dec 08 02:59:57 2007 Dec 08 11:20:41 olli * r9672 /packages/utils/rsync/: [Packages] urtils/sync: One package is enough. We'll keep net/rsync Dec 08 11:21:15 olli * r9673 /packages/net/rsync/Makefile: [Packages] net/rsync: More fixes Dec 08 15:52:29 pavlov * r9674 /trunk/target/linux/generic-2.6/ (5 files in 2 dirs): updated layer7 to 2.15.. current is 2.16.1, i might bump again, this has been uncommitted for a bit Dec 08 15:53:07 pavlov * r9675 /trunk/target/linux/x86/Makefile: bump kernel to 2.6.23.9 for x86 Dec 08 15:56:06 pavlov * r9676 /packages/net/l7-protocols/Makefile: update l7 protocols to 11-22-2007 Dec 08 16:58:50 olli * r9677 /packages/utils/lcd4linux/Makefile: [Packages] utils/lcd4linux: Fix compilation Dec 08 19:17:14 juhosg * r9678 /packages/libs/libdaemon/Makefile: [packages] libs/libdaemon: update to version 0.12 Dec 08 19:20:35 juhosg * r9679 /packages/libs/avahi/Makefile: [packages] libs/avahi: update to version 0.6.21 (closes #2729) Dec 08 20:26:19 nbd * r9680 /packages/net/miau/Makefile: Dec 08 20:26:19 Updated miau to 0.6.5, closes #2812 Dec 08 20:26:19 Signed-off-by: Andy Boyett Dec 08 20:36:01 Is there any way to add a comment to a quilt-managed patch? - so that for example, if you do 'quilt series' you can get informed "this patch should be removed when the kernel is upgraded to 2.6.24" Dec 08 20:36:15 Would anybody find it useful? Dec 08 20:36:45 I have no input until I start using quilt. Dec 08 20:36:55 Fair enough! Dec 08 20:38:03 I embraced quilt a short while ago (and have been pleased). Now it seems I'm still playing catchup, as nbd et al. have started using git heavily! Dec 08 20:38:14 nbd, r9679 dg834gv1 dmesg http://openwrt.pastebin.com/m71291374 logread http://openwrt.pastebin.com/m65948d9d Dec 08 20:38:38 git is the new subversion Dec 08 20:39:08 Indeed - and I thought I was doing well having just about mastered subversion Dec 08 20:40:26 noz: hehe, I finally convinced a CVS user in the office to switch to svn, and roughly three weeks after that I jumped to git. Dec 08 20:40:48 (he's been using CVS roughly since its inception) Dec 08 20:42:01 I rolled out subversion to work about 2 1/2 years ago. They were using CVS. I was new to both. We've had about 15000 commits so far, and it's been fairly smooth. Dec 08 20:44:27 noz: the nice thing about git is that it functions quite well as a svn client. Dec 08 20:44:43 it can even function as a cvs server :-) Dec 08 20:46:59 :) I know I should look into it. I suppose it really depends on your development model. If you pass patches backwards and forwards to an "upstream" vendor, then it probably makes sense. If you do mostly local development, and take any upstream mostly verbatim with only locally-applicable mods, then it probably doesn't make much sense to switch to git. If you're heavily collaborating, I can see it makes a lot of sense. Dec 08 20:49:18 noz: git merging is far superior to svn, it makes integrating new upstream versions much simpler. Dec 08 20:51:15 Oh. Wow. Simpler than 'vendor-import dir module old new', 'cd work', 'svn merge $SVNROOT/vendor/module/old $SNVROOT/vendor/module/new'? Then I must look at it. Dec 08 20:52:00 (OK, so we've done a little local work on a vendor-import script) Dec 08 20:52:40 But I *do* like the idea of being able to cherry-pick patches... Dec 08 20:53:43 noz: the feature that sold me on it is git bisect Dec 08 20:54:22 it provides an automated bisection algorithm to determine which commit introduced a problem Dec 08 20:54:22 Seen a bit of discussion about that. What does it do? Dec 08 20:54:52 define a known good revision in the past, and a known bad revision Dec 08 20:55:06 ... but presumably, you have to try out each intermediate? It just tells you which ones to try? Dec 08 20:55:10 it splits the diff down the middle, checkouts that revision Dec 08 20:55:36 I guess it's cool having tool help for that, yes Dec 08 20:55:39 work with it, determine whether it is good or bad, then tell git with "git bisect good" (or bad) Dec 08 20:56:00 splits the remaining revisions down the middle and repeats. Dec 08 20:57:10 if you write a script that does the tests you need to confirm the problem you can use that via "git bisect run