[01:06] <ebroder> ScottK: is it reasonable to INVALID a backport request for a bug fix? There's a separate bug tracking an SRU for the same fix
[01:07] <ebroder> (bug #666204; sru request is bug #619015)
[01:00] <ebroder> slangasek: i've been just pressing escape as soon as plymouth shows up to see the warning. more reliable than finding the right planetary alignment to see it :)
[02:12] <Kaleo> hi
[02:13] <Kaleo> there is a bug in maverick fixed in the natty package and I cannot find the documentation for the process of requesting a backport of the fix
[02:13] <Kaleo> here is the bug: https://bugs.launchpad.net/ubuntu/+source/xdg-utils/+bug/664531
[02:21] <ebroder> !sru | Kaleo
[02:22] <Kaleo> ebroder: thank you very much
[02:22] <ebroder> np
[07:00] <slangasek> ebroder: oh; feel free to amend the test case
[09:25] <BlackZ> lucidfox: bug #608159: are you working on that merge?
[09:53] <lucidfox> BlackZ> no
[09:53] <BlackZ> lucidfox: ok, if you don't mind, I'd merge it :)
[13:14] <corecode> hey
[13:14] <corecode> i'm trying to package a setuid binary, but somehow that doesn't work
[13:14] <corecode> is there something special i need to do?
[13:14] <corecode> the permissions appear unchanged, already in the staging area
[13:18] <debfx> corecode: dh_fixperms might remove the setuid bit
[13:19] <corecode> ah, thanks
[13:19] <corecode> that's it
[13:20] <corecode> or not :/
[13:21] <corecode> ah yes
[15:00] <dylan-m> d
[15:01] <iulian> e
[15:01] <BUGabundo> f
[15:03] <azeem> dylan-m: please don't do that
[15:53] <corecode> any suggestions how to run a single command on bootup from a package?
[15:53] <corecode> provide an init script?
[15:53] <corecode> or add a @reboot line to cron?
[15:59] <debfx> corecode: I'd go with an init script or upstart job
[16:01] <corecode> i'd like to keep it compatible with debian
[16:01] <corecode> do they use upstart?
[16:01] <debfx> no
[16:01] <corecode> ok
[16:05] <debfx> don't forget to check in the init script if the package is still installed
[16:18] <ggeorgy> hi
[16:44] <highvoltage> persia: debian has yet another definition for 'flavour': http://wiki.debian.org/DebianPureBlends
[16:45] <bgamari> Scott Remnant in the house?
[16:47] <highvoltage> bgamari: he's Keybuk on irc, so doesn't seem so
[16:51] <bgamari> Is it common that distribution specific changes to kernel parameters exposed through /proc/sys are changed in the kernel tree instead of /etc/sysctl.conf?
[16:51] <bgamari> Scott seems to think that if a change in the default swappiness were to happen it should happen in the kernel tree
[16:52] <bgamari> which seems a little strange to me
[16:52] <bgamari> isn't this the precise reason we have sysctl(8)
[17:00] <ogra_ac> bgamari, well, not if you can convince upstream to take the change
[17:00] <ogra_ac> its one piece less to maintin
[17:00] <ogra_ac> *maintain
[17:01] <ogra_ac> sysctl is to override defaults
[17:05] <bgamari> ogra_ac: Precisely
[17:05] <bgamari> ogra_ac: Upstream will not take a change in the default swapiness
[17:05] <bgamari> this seems quite clear
[17:06] <bgamari> especially not a change to what ubuntu should be using in its desktop distribution (probably between 10 and 30)
[17:09] <bgamari> I suspect Scott was suggesting that the swapiness should be changed in the ubuntu kernel tree
[17:10] <bgamari> but still, as you said, sysctl is to override defaults
[17:10] <bgamari> why would we pollute the ubuntu tree with patches that have no hope of making it upstream when there is a perfectly good mechanism for overriding this value in userspace?
[17:11] <ebroder> sysctl can also be used to set defaults. c.f. /etc/sysctl.d/10-zeropage.conf
[17:12] <ebroder> err, reading the comments, i guess that's actually a bad example
[17:12] <bgamari> Either way, this insistence on going through the kernel tree is one of the few reasons #516834 hasn't moved forward
[17:14] <bgamari> we have known that the current swapiness is ill-suited for desktop use for almost a year
[17:14] <bgamari> s/known/recognized/
[17:14] <bgamari> it's been known for far longer than that
[17:16] <ebroder> reading the discussion, i agree with Keybuk. and it looks like the reason the bug isn't making any progress is because people were being stubbornly unhelpful while JFo was trying to triage
[17:17] <ebroder> (primarily that the different kernel flavors give you the distinction you want to figure out what swappiness should be set to, which you don't have in the procps package)
[17:19] <bgamari> ebroder: Perhaps there could be packages containing bits of server- or desktop-specific configuration (e.g. sysctl.d files)
[17:19] <ebroder> we have that already. linux-image-generic vs. linux-image-server
[17:19] <bgamari> true
[17:19] <bgamari> I guess that would be the appropriate place in that case
[17:24] <bgamari> ebroder: It's true that people weren't particularly helpful
[17:24] <bgamari> but it's also true that a vague request for "testing" with no further qualification isn't going to get you particularly useful responses
[17:26] <ebroder> bgamari: people were actively unhelpful and antagonistic to the guy whose job it is to figure out what bugs the kernel team should be paying attention to. i think the result was fairly predictable
[17:27] <bgamari> fair enough
[17:29] <bgamari> ebroder: The problem with dropping a file in sysctl.d from linux-image-* is that there is no dependency on procps
[17:30] <bgamari> ebroder: Is it acceptable to carry a file in package A that has no use without package B, while there is no relationship between A and B
[17:32] <ebroder> is there not a config flag or something to set the default swappiness?
[17:33] <bgamari> let me check
[17:33] <ebroder> hmm...looks like not. that might be a good resolution - come up with a CONFIG_SWAPPINESS or whatever, because that's easy to frob on a per-flavor basis
[17:33] <bgamari> doesn't look like
[17:33] <bgamari> it
[17:34] <ebroder> and that might be upstreamable
[17:34] <bgamari> definitely
[17:37] <bgamari> I'll try working up a patch
[17:56] <bgamari> ebroder: patch attached
[17:58] <ebroder> bgamari: i'm not really qualified to pass judgement, but that looks reasonable to me
[17:59] <ebroder> should CONFIG_DEFAULT_SWAPPINESS depend on CONFIG_SWAP or something like that?
[17:59] <bgamari> probably
[18:00] <ebroder> (i don't really have a good understanding of how kconfig works)
[18:00] <bgamari> I'll fold that in and send it out as an RFC to see how repulsive the kernel folks find it
[18:00]  * ebroder nods
[18:00] <bgamari> I don't have great hope
[18:01] <bgamari> but you never know
[18:01] <maco> hi ben
[18:01] <bgamari> by this logic, the kernel should expose all sysctl knobs in Kconfig
[18:01] <bgamari> maco, hey!
[18:01] <bgamari> long time
[18:01] <maco> aye! get back down to dc for a visit! :P
[18:02] <bgamari> heh, we'll see what happens
[18:02] <bgamari> First I have to pass my quals up here
[18:02] <bgamari> I probably shouldn't even be doing coding work at the moment
[18:02] <bgamari> but a man can only take so much physics
[19:29]  * sanchaz brb
[21:14] <debfx> ScottK: what's blocking no-change backports? it's been over a month since you approved gcc-3.3 and virtualbox