[06:35] <dholbach> good morning
[07:04] <geser> good morning dholbach
[07:04] <dholbach> hi geser
[08:15] <Laney> hey
[08:15] <Laney> dholbach: aren't you abroad somewhere?!
[08:16] <dholbach> hey Laney
[08:16] <dholbach> I'm at work :)
[08:20] <Laney> :P
[08:48] <yolanda> hi, can i get some help with a package? i'm having a problem with postgresql dependencies
[08:49] <yolanda> i have a bug reported for that: https://bugs.launchpad.net/ubuntu/+source/openerp6.1/+bug/1061418
[08:49] <yolanda> yes, ubottu, that's it
[08:50] <yolanda> i have the problem that i need to execute a command to create a postgresql user
[08:50] <yolanda> but as postgres is listed as recommended, it's installed after the openerp package, so the command isn't valid if postgres isn't already there
[08:51] <yolanda> we wanted to list postgres as recommends because if people chosee remote postgres connection, the package isn't needed, but it's needed with local connection... so i'm lost at this point
[08:58] <yolanda> hi, can i get some help with that bug? https://bugs.launchpad.net/ubuntu/+source/openerp6.1/+bug/1061418
[09:01] <geser> you can't really do much about it. you could provide a script to help the admin to set up the database (be it either local or remote)
[09:04] <yolanda> geser, at the moment it configures correctly postgres, but only if postgres is installed on the system, can i force the installation before openerp package?
[09:05] <Daviey> It sounds to me suitable Recommends on postgresql not being set, or on the correct package..  but i haven't checked.
[09:06] <Laney> yolanda: if you want to use stuff in your maintainer scripts, you need to have it as Depends
[09:06] <Laney> s/maintainer scripts/postinst/
[09:07] <yolanda> Laney, i have it as recommends, so i should change to depends
[09:07] <Laney> that's how you guarantee to have it available at the right time
[09:08] <yolanda> but for some cases postgres isn't needed, so that was my doubt
[09:08] <Laney> well, then why do you always need to run postgres stuff in the postinst?
[09:08] <yolanda> Laney, the thing is that i let the user to choose between local and remote installation
[09:08] <yolanda> if the installation is remote, postgres isn't needed
[09:08] <yolanda> but it installation is local, it's needed and i need to run an script
[09:09] <Laney> perhaps you should print an informational message to the user if you can't run the setup then
[09:09] <Laney> or some other solution like have openerp-local openerp-remote packages with different scripts
[09:10] <Daviey> doing a [ -f /usr/sbin/postgres ] setupdb() .. seems OK .. but cleaner might be openerp-standalone .. which is a meta-like-package (depends on current stuff, and postgres) with the maintainerscript to do the db setup?
[09:10] <yolanda> Laney, but i already have the package split in two, between openerp-core that provides only the package and openerp services, and openerp-full, that configures the database connection and leaves openerp ready to run
[09:11] <yolanda> openerp-core is supposed to be used by advanced users, and openerp-full for general users
[09:11] <Daviey> (then full needs Depends?)
[09:12] <Laney> so perhaps -core could automatically do the remote configuaration, then -full reconfigures this to be local?
[09:12] <yolanda> Daviey, Laney, the question between remote and local is on the full package, at the core we still don't know about the type of connection
[09:13] <Laney> we're suggesting making some assumptions
[09:13] <Laney> that the user can override
[09:14] <Laney> (but not at package installation time)
[09:15] <yolanda> Laney, don't understand
[09:15] <yolanda> can i address you to the source code for the package so you can check what i've done and if you detect something that can be improved?
[09:18] <Laney> yolanda: I'm saying that if you want to make it unconditional then you need the Depends, and then this raises some questions about the local/remote split
[09:18] <Laney> you should consider what you think is the best approach based on this
[09:19] <yolanda> Laney, i'm tempted to just add a depends, but it's one of my first packages, so i can't have a clear opinion on it
[09:19] <cjwatson> Or a conditional self-trigger might be possible, but it's kind of complex
[09:20] <cjwatson> (And runs the risk of an orphaned trigger if postgres never gets installed, so I think on reflection that's probably not a good idea)
[09:22] <yolanda> ok ,so simplest way is to just add a Depends, and the worse consequence there is that postgres is installed and not used, right?
[09:23] <Laney> then users who don't need it won't be able to uninstall it
[09:24] <Daviey> well -full is the full shebang.  So not necessarily a problem
[09:24] <geser> yolanda: what's the difference between -core and -full? just the automatic database setup?
[09:24] <Laney> I'm not saying that it is or isn't, just presenting the facts
[09:25] <yolanda> geser, full configures the postgres user, and adjusts settings in openerp configuration file
[09:25] <Laney> so -core is essentially the remote configuration anyway?
[09:25] <semvoz> good evening
[09:31] <yolanda> Laney, no, core just installs the openerp source code in the right places, and installs the openerp daemon
[09:31] <yolanda> then, advanced users could configure openerp manually
[09:32] <yolanda> but full is intented to be for basic users, so they just get openerp ready to go
[09:32] <yolanda> i think a depends make sense, because normally basic users will choose local configuration
[10:31] <semvoz> hmm
[13:13] <ara> micahg, ping
[14:29] <TheLordOfTime> anyone here on bugcontrol want to remind me of how we handle bugs against EOL-release packages in general, for packages which don't fall into a specific workflow?
[14:29] <TheLordOfTime> (i'm not getting answers in -bugs, but i know there's bugcontrollers in this channel :P)
[14:41] <cjwatson> TheLordOfTime: unless there's some particular reason why the bug is specific to that release, there's likely nothing special; if it's not been closed it probably still exists
[14:42] <TheLordOfTime> cjwatson, all i'm seeing is crap in 10.10, but since this is in a server package, hallyn already answered.
[14:42] <cjwatson> TheLordOfTime: in the case of something that's basically a support request or in some other way not well-formed as a bug, it might be a good time to encourage the reporter to recheck
[14:42] <TheLordOfTime> typically my response is to say "Please test in $nonEOLRelease to confirm if the bug exists"
[14:42] <cjwatson> TheLordOfTime: but if it's a well-formed bug report, that just wastes the reporter's time and makes us look foolish
[14:42] <cjwatson> TheLordOfTime: it's the number one thing my friends complain about when they've filed bugs in Ubuntu
[14:43] <TheLordOfTime> cjwatson, tell me about it
[14:43] <TheLordOfTime> i see that in nginx packages, and i rage about it :P
[14:43] <cjwatson> not that their bugs don't get fixed, but that somebody who didn't understand their bug came along, marked their bug incomplete, and asked them if it still existed, even though there was a perfectly good reproduction recipe in the bug
[14:43] <cjwatson> so yeah, try to make an effort to actually understand the bug before essentially auto-posting to it
[14:43] <TheLordOfTime> cjwatson, if you're curious, https://bugs.launchpad.net/ubuntu/+source/php5/+bug/744371 is the bug, but hallyn's given me a method of how to handle it
[14:44] <cjwatson> I'm not curious :)
[14:44] <TheLordOfTime> cjwatson, with no response since 2011-09-08, i'm assuming its not been tested in later releases :P
[14:44] <TheLordOfTime> and that its just fallen off the radar
[14:44] <cjwatson> that's in general an unreasonable assumption
[14:45] <cjwatson> how about actually testing
[14:45] <TheLordOfTime> cjwatson, works in 12.04
[14:45] <TheLordOfTime> and worked in 11.04 when i ran that
[14:45] <TheLordOfTime> 11.10 too, but in 11.04 and 11.10 i dont have any existing testbeds
[14:45] <TheLordOfTime> since those systems were retired in favor of Precise
[14:46] <cjwatson> then your setup apparently doesn't match, since hallyn reported that this was still the case in natty
[14:46] <cjwatson> a good test would involve a system that could reproduce the bug being upgraded and now no longer reproducing the bug
[14:46] <cjwatson> rather than one that could never reproduce it in the first place
[14:48] <TheLordOfTime> cjwatson, ah, i see that now, hallyn didnt mention that :P
[14:48]  * TheLordOfTime yawns
[14:48] <TheLordOfTime> being tired results in minor oversights like that
[14:48] <TheLordOfTime> although, natty's EOL this month, i'd like to see it tested in 11.10 :P