xnox | Chipaca: please use /etc/machine-id, anything else is broken. | 00:00 |
---|---|---|
slangasek | xnox: nack | 00:00 |
slangasek | /etc/machine-id is *also* broken | 00:00 |
slangasek | it's *the same on every device* | 00:00 |
Chipaca | xnox: for what? | 00:00 |
xnox | slangasek: WHAT?! | 00:00 |
slangasek | you're jumping the gun slightly :P | 00:00 |
slangasek | xnox: it's in the base image, how would it have been made unique? | 00:00 |
xnox | slangasek: /etc/machine-id must not be the same = | 00:00 |
slangasek | that's nice ;) | 00:00 |
xnox | Chipaca: push & unique identification..... | 00:01 |
xnox | slangasek: let me check livefs image to see if all /etc/machine-ids are the same on all ubuntus.... | 00:01 |
xnox | (ditto dbus ids) | 00:01 |
slangasek | dbus ID is handled | 00:02 |
slangasek | /etc/machine-id is a neologism | 00:02 |
xnox | slangasek: /var/lib/dbus/machine-id should be a symlink to /etc/machine-id in the new world order.... | 00:02 |
slangasek | all these opinions | 00:02 |
slangasek | fwiw mine are in sync, but it's not a symlink | 00:02 |
slangasek | and I don't see what creates /etc/machine-id | 00:03 |
xnox | slangasek: ok. i have one task to fix, i should focus on that =)))) | 00:03 |
xnox | *sigh* /etc/machine-id is on the livefs image.... | 00:04 |
slangasek | ok. what creates it? | 00:04 |
xnox | slangasek: looking at the source code systemd-ish something does it. | 00:04 |
slangasek | just because systemd upstream has defined this as an interface for a unique ID in no way means we've started supporting it as such | 00:05 |
xnox | systemd-machine-id-setup | 00:05 |
slangasek | we still have /var/lib/dbus/machine-id as uniqueu | 00:05 |
xnox | it uses dbus/machine-id as seed, if there was one. | 00:06 |
slangasek | xnox: so in the long term, the installer / live-builder should be updated to handle /etc/machine-id specially, just as /var/lib/dbus/machine-id is currently handled. In the near term, /etc/machine-id is an irrelevant and unsupported interface. :) | 00:11 |
xnox | slangasek: bug against cdimage project? | 00:11 |
slangasek | xnox: I still have no idea what is meant to be creating /etc/machine-id, 'systemd-machine-id-setup' isn't part of either the boot scripts or the maintainer scripts | 00:12 |
slangasek | xnox: yes | 00:12 |
xnox | slangasek: installer is suppose to call systemd-machine-id-setup, otherwise, i guess first caller triggers it's creation.... | 00:13 |
slangasek | what's the "first caller"? | 00:13 |
slangasek | do you mean the first thing that queries the machine ID? | 00:14 |
xnox | systemd-firstboot.service , but of course /me is friday trolling | 00:14 |
xnox | i'm guessing logind session. | 00:14 |
xnox | or same as dbus-uuigen. I'll trace that later. No clue yet, where/how it ends up generated on ubuntu. | 00:15 |
juliank | I'd recommend to ask systemd guys or look at more recent systemd code. Given all that booting without /etc stuff they're doing, they probably have newer versions generate that if it is missing | 00:17 |
juliank | Might make sense to copy that then | 00:18 |
hallyn | xnox: uh oh, i notice that the 0.28-3ubuntu1 cgmanager upstart job has the post-start "initctl ... || true" in exec instead of inside script. Needs an update? | 00:23 |
xnox | hallyn: no, why? | 00:23 |
hallyn | oh, i thought there was talk last week of || true not working in a exec | 00:23 |
hallyn | if not, excellent | 00:24 |
xnox | hallyn: exec -> single line; script / end-script -> multi-line | 00:24 |
xnox | hallyn: if upstart spots any special / shell characters in the exec line it runs it through a shell. | 00:24 |
xnox | hallyn: i'd like to see it to believe it =) we have extensive tests to make sure e.g. "exec daemond --bar $ARGS" ends up called through a shell. | 00:25 |
xnox | to do all the shell variable expansions and magic. | 00:25 |
hallyn | xnox: right, i thought in particular "exec somethingorother || true" was what wasn't going to work. happy to learn i'm wrong. (had not tested) | 00:27 |
calebjosue | i was recommended to join this channel for purposes of knowing where to start with open source projects | 00:29 |
xnox | hallyn: hm.... I'm scratching my head now. | 00:32 |
xnox | slangasek: hallyn: This fails to start $ echo "pre-start exec false || true" > ~/.config/upstart/hallyn.conf; start hallyn | 00:32 |
xnox | but it should..... | 00:32 |
slangasek | xnox: eh? why should that successfully start instead of being a parse error? | 00:45 |
slangasek | xnox: 'pre-start exec false || true' says "exec() false with the arguments '||' and 'true'" | 00:45 |
slangasek | so technically not a parse error, but certainly returns non-zero and should fail | 00:46 |
xnox | slangasek: "Any special characters, e.g. quotes or `$' specified will result in the entire command being passed to a shell for expansion." imho || is special | 00:46 |
slangasek | xnox: upstart does its own shell parsing of variables on the exec commandline; I don't see any reason '||' should be regarded as special | 00:47 |
xnox | hallyn: i am wrong, and cgmanager jobs need fixing. | 00:48 |
xnox | slangasek: about whoopsie id being different for non-root users -> Bug #1351519 | 00:58 |
ubottu | bug 1351519 in whoopsie (Ubuntu) "/sys/class/dmi/id/product_uuid is not world readable" [Undecided,New] https://launchpad.net/bugs/1351519 | 00:58 |
slangasek | xnox: ok; that's one reason why the generated id should be cached to the filesystem | 00:59 |
xnox | slangasek: and then make whoopsie file recoverable errors upon itself when generated id doesn't match cached one?! =) | 01:00 |
slangasek | ... | 01:00 |
* xnox is still friday trolling | 01:00 | |
xnox | slangasek: e.g. tedg files recoverable errors against upstart, when something rather in url-dispatcher (tedg) can't read an sql database.... | 01:00 |
slangasek | yes, one wonders at your choice of leisure activities | 01:00 |
xnox | slangasek: i'd rather tedg stop spaming us =) | 01:01 |
slangasek | xnox: I've filed a bug against url-dispatcher for this | 01:01 |
slangasek | bug #1350536 | 01:01 |
ubottu | bug 1350536 in url-dispatcher (Ubuntu) "url-dispatcher report_recoverable_problem() handler triggers reports filed against upstart" [Undecided,New] https://launchpad.net/bugs/1350536 | 01:01 |
xnox | slangasek: number? i'll fix it, cause i believe we shouldn't be generating errors on continues basis. All error reports must be exceptional, not a norm. | 01:02 |
slangasek | well, my bug report was about the errors being misfiled | 01:02 |
slangasek | which is separate from whether there's a bug in url-dispatcher /triggering/ the problem report | 01:02 |
=== timrc is now known as timrc-afk | ||
hallyn | xnox: ok, thx, i'll push a fix (but not right now, laptop overheating) | 01:45 |
hallyn | xnox: slangasek: so how many upstartu sers do we think we have in debian? should the upstart job fix for cgmanager go to debian or just to utopic? | 01:47 |
slangasek | is there some reason to *not* fix it in Debian? to do otherwise means you get to maintain a delta | 01:48 |
xnox | hallyn: all cgmanager updates should go to debain, and just sync to ubuntu. | 01:48 |
xnox | (e.g. that's how a lot of core packages are maintained e.g. grub2 is synced from debian, all uploads go into debian) | 01:49 |
hallyn | xnox: yeah pitti had made an ubuntu delta this morning, but i do think it can a go to debian | 01:50 |
hallyn | i.e. it was a delta that was introduced while 0.28-3 was going into debian, and it failed ot make it into that package (i didn't track them all sufficiently) | 01:51 |
hallyn | unfortunately going through debian adds a delay | 01:52 |
slangasek | hallyn: yes, but it's not something that should be a delta between the two distros over the long term | 01:53 |
hallyn | slangasek: no, not at all - it's just a delta bc i failed to put it into 0.28-3 | 01:54 |
xnox | Folks please tell me: | 05:14 |
xnox | $ adb shell cat /sys/class/android_usb/android0/iSerial | 05:14 |
xnox | Need some sample data, to validate a hypothesis =) | 05:14 |
=== desrt|guadec is now known as desrt | ||
=== timrc-afk is now known as timrc | ||
=== timrc-afk is now known as timrc | ||
ScottK | zul: Please use build1 not ubuntu1 when you're doing no change rebuilds of packages that don't have an Ubuntu diff already. | 19:19 |
Noskcaj | Could someone please work on porting ubuntu-system-settings, powerd, and indicator-power to upower 0.99 soon? | 20:54 |
Noskcaj | Laney, Since you show up in the changelogs a bit do you know if anyone's working on ^ | 20:56 |
slangasek | Noskcaj: why is porting required at all? Shouldn't upower be a stable interface already? | 21:08 |
Noskcaj | slangasek, The new (late last year) release of upower drops the suspend API, the SONAME, and changes part of the dbus interface, as well as a few other things. | 21:12 |
slangasek | "drops the suspend API" in favor of what? | 21:12 |
Noskcaj | gnome, and soon xfce and kde, needs upower 0.99 to function | 21:12 |
Noskcaj | slangasek, systemd/logind | 21:13 |
slangasek | ok | 21:13 |
Noskcaj | bug 1330037 | 21:13 |
ubottu | bug 1330037 in xfce4-settings (Ubuntu) "upower 0.99 transition" [Undecided,In progress] https://launchpad.net/bugs/1330037 | 21:13 |
slangasek | and do you know that powerd actually requires porting for any of these changes, vs. a straight rebuild? | 21:13 |
slangasek | (ditto u-s-s and indicator-power) | 21:14 |
Noskcaj | powerd failed to build last i tried, i'll rebuild those three with their current versions today | 21:15 |
Noskcaj | (ditto u-s-s and indicator-power) | 21:15 |
Noskcaj | crap | 21:15 |
Noskcaj | https://launchpadlibrarian.net/177594528/buildlog_ubuntu-utopic-amd64.powerd_0.15%2B14.10.20140612-0ubuntu1ppa1_FAILEDTOBUILD.txt.gz | 21:15 |
slangasek | hmm, ok then | 21:20 |
slangasek | I'm guessing that for powerd porting you might want sforshee | 21:21 |
Noskcaj | powerd is still failing porting you | 21:30 |
Noskcaj | https://launchpad.net/~noskcaj/+archive/ubuntu/upower/+build/6235759/+files/buildlog_ubuntu-utopic-amd64.powerd_0.16%2B14.10.20140707-0ubuntu1ppa1_FAILEDTOBUILD.txt.gz | 21:31 |
Noskcaj | Why does hexchat copy anything i select :( | 21:31 |
Noskcaj | indicator-power does build fine, but it probable needs some testing, as it does depend on upower | 21:36 |
* slangasek nods | 21:37 | |
Noskcaj | u-s-s https://launchpad.net/~noskcaj/+archive/ubuntu/upower/+build/6235765/+files/buildlog_ubuntu-utopic-amd64.ubuntu-system-settings_0.3%2B14.10.20140801.2-0ubuntu1ppa1_FAILEDTOBUILD.txt.gz | 21:38 |
slangasek | note that, due to the upcoming release-to-manufacturer milestone for Ubuntu Phone, it's likely to at least be a week before anyone is willing to work on this for utopic (basically because we aren't going to want to pull upower 0.99 into the RTM archive) | 21:39 |
Noskcaj | ok | 21:40 |
Noskcaj | pitti, Any progress on python-dbusmock upower? | 21:40 |
Noskcaj | And once this is done, we get the fun of bluez5 | 21:41 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!