[04:34] <catid> hello.. the respawn COMMAND syntax no longer works right?
[04:35] <catid> i'm trying to update my svscan script in event.d to work with Fiesty
[04:35] <catid> respawn exec /command/svscanboot doesn't seem to do the trick
[04:36] <catid> would someone please help me to craft this so upstart will do the right thing?
[04:40] <catid> eh, just removing respawn from the line works..
[04:40] <catid> having it respawn would be nice though
[04:41] <catid> later..
[09:44] <cwillu> is upstart broken'ish on edgy?
[09:57] <sadleder> cwillu: why do you think so?
[10:02] <cwillu> more likely my assumptions are off
[10:02] <cwillu> events/jobs are run as root?
[10:03] <cwillu> I can't seem to get any output out of a script in an exec or respawn line;  I know the script runs because the process stays alive if I change the script to 'sleep 100m', but it seems like if I put anything else in there, it doesn't do anything.  It seems to work fine if I run it via sudo though
[10:15] <cwillu> ah, no $HOME
[10:15] <cwillu> this is why you don't do work at this hour
[10:15] <cwillu> while the drunks are still in the office (grr)
[10:19] <sadleder> hmm, was that the problem?
[11:05] <sadlede1> cwillu: could you resolve the problem?
[11:12] <cwillu> ya, environment wasn't as I expected
[11:12] <cwillu> there's still a udev glitch on edgy that I'm working around, but things are working
[11:18] <sadlede1> cwillu: ah, ok
[11:19] <sadlede1> cwillu: for what task are you using upstart on edgy?
[11:20] <cwillu> multiseat autoconfig;  udev to assign keyboards devices names corresponding to existing displays, and then triggering an upstart event to launch an evdev bridge from that device to the actual display
[11:20] <cwillu> the idea being that a keyboard should be able to be unplugged, and a different one plugged in, and it'll get assigned to the display that's missing a keyboard
[11:21] <sadlede1> wow
[11:21] <cwillu> additionally, if a display used a particular usb port before, it'll prefer to use a keyboard on that port again if available
[11:22] <cwillu> which makes the initial setup much easier:  don' t have to move keyboards around if they don't go to the right monitors when it's first set up, you can just swap the plugs
[11:23] <cwillu> from there on in, as long as you don't unplug more than two keyboards at the same time, and then plug them into physically different ports, everything will work as expected
[11:23] <cwillu> important when somewhat computer illerate people will need to keep 5 keyboards pointing at the right 5 monitors :p
[11:23] <cwillu> (re:  said drunks)
[11:24] <cwillu> only thing left is to selectively re-trigger the udev event for keyboards that are unassigned when a new display comes up
[11:26] <cwillu> I figure it should save us about $10,000 a year in electricity alone (16 less computers, better options for power savings on the ones that are left)
[11:27] <sadlede1> what is the system you are developing this for?
[11:27] <cwillu> phone order stations for a pizza shop
[11:27] <cwillu> currently use a bunch of old machines, one per phone line
[11:30] <sadlede1> cwillu: sounds like a great project
[11:52] <sadlede1> cwillu: so you have 5 graphic adapters?
[11:52] <sadlede1> in one machine?
[11:52] <cwillu> 3 dual headed
[11:53] <cwillu> although I'm planning on using cheaper singleheaded cards in other machines
[11:53] <cwillu> there's another bit of magic to make this work on sharedfb displays :p
[04:19] <_ion> Heh
[04:20] <Keybuk> what amuses me is mostly how many inches of column can be dedicated to a licence that doesn't even exist yet
[04:21] <_ion> :-)
[04:22] <Keybuk> I've had maybe a dozen e-mails so far asking whether Upstart will be GPL-3
[04:23] <_ion> Heh, that's interesting
[04:23] <Keybuk> I just reply that I can't make any decision until I've read the final version of the licence, and until then it will definitely remain GPL-2
[04:26] <_ion> The source can be used under the terms of GPL-3 as well, as soon as it exists anyway.
[04:26] <_ion> unless you remove the "or (at your option) any later version" clause. :-)
[04:26] <Keybuk> sure, I don't see a problem there
[04:27] <Keybuk> the FSF have always promised the GPL-3 will only be more restrictive
[04:27] <_ion> Yeah
[04:27] <Keybuk> it won't allow people to do things the GPL-2 prevents people from doing
[04:27] <Keybuk> so someone has the option of modifying and/or distributing the source /under the terms/ of the later version
[04:27] <Keybuk> they don't have permission to *change* the terms, just follow the more restrictive ones
[04:27] <_ion> Indeed.
[04:28] <Keybuk> and that permits the code being combined with GPL-3 code at a later date
[04:29] <Keybuk> providing the GPL-3 isn't bad, there's no problem there
[04:29] <Keybuk> if there's real problems with it, I'll remove the clause :p
[04:39] <_ion> Etsivt etsivt etsivt etsivt etsivt etsivt etsivt.
[04:39] <_ion> That translates roughly to "the private investigators, that search for the private investigators, search for the private investigators that search for the private investigators". :-)
[04:40] <Keybuk> ?#
[04:41] <_ion> Something akin to the http://en.wikipedia.org/wiki/Buffalo_buffalo_Buffalo_buffalo_buffalo_buffalo_Buffalo_buffalo thing :-)
[05:18] <cortana> nice link
[05:52] <cwillu> what version of upstart supports arguments on events?
[05:52] <Keybuk> 0.3.2 I think
[05:52] <Keybuk> 0.3.8 is the current recommended version
[05:53] <cwillu> :/  edgy is runs 2.something I think
[05:53] <cwillu> no environment vars either, huh
[05:53] <_ion> Feisty will be released soon. :-)
[05:53] <cwillu> this is true, which just means I have to take out whatever hack job I use to get this working this month when its release next month :p
[05:54] <cwillu> know of any good hackjobs to do this?
[05:54] <_ion> What are you trying to do?
[05:55] <cwillu> I've got some udev stuff that maps keyboards to heads on a multi-seat system, but doesn't actually do the connection (just names them after the server that should use them)
[05:55] <cwillu> I've got another program that reads the keyboard and forwards the events via xtest
[05:56] <cwillu> I want to have that last program launched via upstart
[05:57] <cwillu> "initctl trigger kbd/added <device name>" is what I'm after
[05:58] <cwillu> but there may be anywhere from 1 to 12 heads, so I'd like to avoid making 12 upstart event definitions if possible
[05:59] <Keybuk> you want to use the current version of upstart, then
[05:59] <cwillu> that's what I hear :)
[05:59] <cwillu> does it work with edgy?
[06:00] <_ion> It's probably easiest to grab the feisty source package and build it.
[06:01] <_ion> So you get the 0.3.8-compatible Ubuntu jobs as well.
[06:01] <cwillu> let me rephrase:  has the job definition syntax changed enough to break edgy jobs?
[06:01] <cwillu> ahh
[06:02] <cwillu> this is becoming a bit of a beast;  already running fiesty versions of some udev scripts (by-path symlinks otherwise stopped being generated if I plugged in more than a couple keyboards), evince (didn't handle landscape printing in edgy), 
[06:04] <_ion> There's always the possibility of upgrading to feisty already.
[06:05] <cwillu> running it at home, true enough;  haven't run into any issues there
[06:06] <cwillu> though come to think of it, I'm already going through the legwork to make sure the device name is constant, even if the physical keyboard for that display changes.  I could just poll the device
[06:13] <cwillu> would make it an easy fix for fiesty at least
[06:24] <cwillu> is there a handy script to wait for a display to come up?  I can rip the code out of xinit, but if it already exists...
[06:27] <_ion> Can't you just use xinit directly?
[06:30] <cwillu> it doesn't really fit in very well with what I'm trying to do;  would work fine for the nested servers, but the master server, I don't want any window managers running, I don't want it to exit when the clients quit, and the list of clients is generated dynamically
[06:31] <cwillu> I already checked the xinit source code, it's just polling xopendisplay, so I'll probably just reimplement that
[06:32] <cwillu> I dunno; xinit just leaves a bad taste in my mouth
[06:44] <cwillu> and thus was born xpoll :p
[07:03] <cwillu> is there a trick to getting upstart to redetect /etc/event.d scripts under edgy?
[07:04] <Keybuk> it does it by itself
[07:06] <cwillu> so I should be able to dump a one line "exec <some script>", and launch that job immediated via "start <filename in event.d>"?
[07:06] <cwillu> immediately even
[07:06] <Keybuk> yes
[07:07] <cwillu> 'cause I can't :p
[07:07] <Keybuk> oh
[07:07] <cwillu> "initctl: Unknown job: masterServer"
[07:07] <Keybuk> /etc/event.d/masterServer exists ?
[07:07] <cwillu> cwillu@8th:/etc/event.d$ ls -l masterServer 
[07:07] <cwillu> -rw-r--r-- 1 root root 149 2007-03-23 12:05 masterServer
[07:08] <Keybuk> check the syslog to make sure theres no parse error
[07:08] <cwillu> nothing shows up in syslog, although I did have a couple hits on some other jobs I just tried
[07:09] <Keybuk> try opening it and saving it again :p
[07:09] <Keybuk> 0.2 wasn't very good at this
[07:09] <cwillu> I'm noticing this :)
[07:09] <cwillu> okay, got something that time
[07:16] <Keybuk> gnargh
[07:16] <Keybuk> yet another "why doesn't upstart use yacc to parse its config files?" e-mail
[07:17] <Keybuk> 1) because yacc does not generate parsers 
[07:22] <cortana> it doesn't?
[07:25] <Keybuk> no ... it generates a grammar reader
[07:25] <Keybuk> e.g. you can tell bison
[07:25] <Keybuk>   expression: expression '+' expression
[07:25] <Keybuk>             | NUMBER
[07:26] <Keybuk>             ;
[07:26] <Keybuk> but you can't tell it how to extract NUMBER or '+' from a file, or where the whitespace goes, etc.
[07:26] <cortana> well, is that not lex's job
[07:26] <Keybuk> exactly
[07:27] <Keybuk> the correct question would be "why doesn't upstart use LEX/FLEX to parse its config files?"
[07:27] <Keybuk> :p
[07:27] <cortana> ok now i know i didn't totally misunderstand what they o :)
[07:27] <cortana> exactly ;)
[07:27] <Keybuk> yacc/bison aren't much use here, because the config grammar is simpler to deal with using a function table
[07:27] <cortana>  & does it?
[07:27] <Keybuk> no
[07:27] <cortana> but then again lexing is piss easy
[07:27] <cortana> i'm surprised this is such an FAQ