[14:41] I am trying to start upstart from /init and it works but it runs in the foreground and i have no console ... any help is appreciated [14:41] exec upstart 3 [14:42] is the last command in /init [14:43] well i do have a console but is controlled by upstart so it just has upstart messages if i use exec /sbin/init --debug 3 [14:44] last command in /init is "exec /sbin/init --debug 3 [14:45] using ubuntu 9.10 [15:10] right .,.. so i get a console when i ssh into machine and "initctl start tty1". So, why doesnt that just start when i "exec /sbin/init 3" ? [15:12] thaper: what's in your tty job definition? [15:16] 1 sec [15:17] respawn [15:17] exec /sbin/getty -8 38400 tty1 [15:18] start on stopped rc RUNLEVEL=[2345] [15:18] stop on runlevel [!2345] [15:18] not in that order [15:19] i didnt change the tty job definition at all [15:19] which did you change? [15:20] none [15:20] i didnt change any of them [15:21] I am just running "exec /sbin/init 3" from /init ... it is the last command [15:22] when i ssh into the machine ... "initctl list" shows all jobs "stop/waiting" [15:24] when i "initctl start tty1" a login prompt apears on the machine [15:24] idk what /init is [15:25] this is likely a distro-specific question, in which case it would be inappropriate for this channel anyway [15:25] it is just a script [15:26] that runs a few commands that need to be run before /sbin/init [15:26] the kernel jumps there before /sbin/init [15:26] so i start /sbin/init as the last command in /init [15:27] this likely has a lot to do with your distro's deployment of upstart and not upstart proper, and also doesn't sound like the right way of doing whatever it is you're trying to do. [15:28] hmmmmmm it works fine with the sysvinit... [15:28] it works with gentoo sysvuinit [15:29] i need to get this working because i want to run it on any distro that uses upstart [15:29] thank you fr your hep though [15:29] help [15:29] heheheeh [15:30] what is upstart proper ? [15:30] that i would like to know [15:30] do you have a link ? [15:30] thanks sadmac2 [15:33] well i looked into the source and if upstart is not pid1 then telinit runs and then upstart becomes pid1 so there should be no problem with starting /sbin/init from a script [15:33] wonder why it hangs though [16:13] thanks sadmac2 fixed with "start on startup" in tty1.conf [18:11] mbiebl: are you going to Debconf? [18:11] JanC: are you? [18:12] Keybuk: probably not === sadmac2_ is now known as sadmac [18:30] been too long since I conference'd. Hopefully I'll make it to FUDCon F14 [18:31] TOMCAT! [18:31] Keybuk: wtf is tomcat? [18:31] an F-14 [18:32] Keybuk: or a Java app server (*shudder*). What are you getting at? [18:33] I'm just amused that Fedora release numbering now co-incides with US Air Force plane designations [18:33] clearly F14 has to have some Top Gun inspired codename [18:33] Fedora 14 "Maverick" or something [18:35] Keybuk: none of the clever ones like that seem to get through the naming process [18:35] Keybuk: We really wanted 11's code name to be "goes to" [19:03] Keybuk: I'm running a copy of the latest lucid bits on an older laptop [19:04] And during boot I get those kernel messages [19:04] http://paste.debian.net/66811/ [19:04] Disk access is pretty slow [19:05] When I disable ureadahead it seems I can't reproduc these problems [19:05] no idea on that one [19:05] all ureadahead does is trace or read [19:06] I'm not sure if it is really ureadahead or just some weird side effect [19:06] But I never see those problems on my Debian installation [19:07] I hate disks [19:07] I was scared pretty bad when I noticed that, thought my harddisk is dying [19:09] Keybuk: just found this https://bugs.launchpad.net/ubuntu/+source/linux/+bug/285892 [19:09] so it's most likely not an ureadahead bug :-) [19:12] And I do have as SAMSUNG drive [19:13] cscope may have the worst UI ever [20:17] Keybuk: could I get your thoughts on nih_error's design? had some quick ideas for improvement and wanted to make sure I hadn't missed anything important [20:20] let me know your thoughts first ;) [20:20] the only real aim I had was [20:20] - you shouldn't need to initialise an error to catch one [20:20] - you shouldn't need to pass something to a function to get an error back [20:20] - errors should be switch() able [20:22] Keybuk: well, first there's your documentation assertion that you must check the return value of the function to determine if an error was raised. [20:22] Keybuk: a design where you might "check" whether an error was raised introduces its own problems (or at least temptations to create them) [20:23] Keybuk: I actually like a rather radical alternate assertion: Return value is UNDEFINED if an error was raised. [20:24] then you can't do [20:24] if (! function_call (..)) [20:24] { [20:24] /* handle error */ [20:24] } [20:25] Keybuk: no, you're forced to do: [20:25] foo = function_call (..) [20:25] if (got_error) [20:25] ... [20:25] Keybuk: but there's more evil I have in store [20:26] Keybuk: I'd like to be able to do something like this: [20:26] nihe_(function_call (...)) [20:26] /* go on ignoring errors */ [20:26] where nihe_ is a macro that would do: [20:27] ({ ret = function_call (..); [20:27] if (got_error) [20:27] return /* return is undefined, so stuff something here to shut up GCC */ [20:27] ret; }) [20:28] Keybuk: thereby giving us most of try/catch without any stack tricks whatsoever. [20:28] Keybuk: nih_local still works, etc.