| gastag | hi all, can anyone point me to some documentation on the 0.5 series of upstart? | 11:35 |
|---|---|---|
| Keybuk | gastag: there isn't much | 11:36 |
| gastag | ok, so for stanzas definitions I have to look at the code, right? | 11:37 |
| Keybuk | the documentation is deliberately missing to stop people trying to do too much with it | 11:38 |
| gastag | isn't it considered stable? | 11:39 |
| Keybuk | the interfaces are not stable | 11:39 |
| Keybuk | thus the 0. version number | 11:39 |
| Keybuk | instead I've been concentrating on getting the core code stable | 11:39 |
| Keybuk | and the functionality and design we want | 11:39 |
| Keybuk | the current (0.3 and 0.5) series of Upstart are really only intended (and tested) working in a "backwards compatible" mode | 11:40 |
| gastag | I'm currently using 0.3.9, but I'd like to make use of some of the 0.5 features | 11:40 |
| Keybuk | which ones? | 11:48 |
| gastag | Keybuk, mainly states, to be able to define multiple conditions to be true for a job to be started | 13:00 |
| Keybuk | 0.5 doesn't have states | 13:01 |
| gastag | ok, but you can define multiple job status as 'start on started dbus and started gdm' | 13:03 |
| gastag | isn't this in 0.5 ? | 13:03 |
| Keybuk | yes | 13:04 |
| Keybuk | but it doesn't work | 13:04 |
| Keybuk | for example | 13:04 |
| Keybuk | you start dbus | 13:04 |
| Keybuk | nothing happens | 13:04 |
| Keybuk | you start gdm | 13:04 |
| Keybuk | now your job starts as well | 13:04 |
| Keybuk | which is good | 13:04 |
| Keybuk | now you stop gdm | 13:04 |
| Keybuk | your job stops (I assume you have stop on stopping dbus or stopping gdm as well) | 13:04 |
| Keybuk | and you start gdm again (you were just restarting it) | 13:04 |
| Keybuk | your job stays stopped | 13:04 |
| gastag | so this just works the first time you start a job? | 13:06 |
| Keybuk | yes | 13:07 |
| gastag | why this choice? | 13:07 |
| Keybuk | it wasn't a choice, it just didn't work | 13:08 |
| Keybuk | we thought there was a way to make it work, but that didn't pan out | 13:12 |
| Keybuk | so that's the big fix for 0.10, which is out in June | 13:13 |
| gastag | I understand, thanks | 13:14 |
| keesj | in 0.3.9 you can have multiple start on items | 13:31 |
| gastag | keesj, yes but the conditions are in OR, I'd need AND | 13:33 |
| keesj | to solve thisproblem I guess that one would need to make upstart itself keep a in-memory tree of the "state machine" | 13:41 |
| keesj | but even the "and" is a little strange when your tinkg of everything as a event | 13:45 |
| gastag | I agree, but "started" and "stopped" can be thought as sort of states | 13:49 |
| Plouj | hi | 14:00 |
| Plouj | is python needed for running upstart (d-bus bindings) or is it just necessary for the build/compilation of upstart? | 14:01 |
| Keybuk | just the build | 14:06 |
| Plouj | oh, nice | 14:07 |
| Plouj | then there's hope it will fit on my space limited system | 14:07 |
| gastag | sorry, is the failure in oom_adj open failure just a warning in 0.5.1 or it is still a fatal error? | 14:09 |
| Keybuk | still an error | 14:18 |
| gastag | ah ok, so it wasn't a problem in my system. I saw the bug marked as fix released in 0.5.0, but it still happened to me in 0.5.1 | 14:21 |
| Keybuk | what's the bug#? | 14:27 |
| gastag | 259801 | 14:29 |
| gastag | sorry, it's not fixed in upstart | 14:29 |
| Keybuk | didn't think so :p | 14:34 |
| keesj | gastag: started and stopped are kind of states but there are other states like stopping... | 15:43 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!