*** bigbrother has quit IRC
*** curt has quit IRC
*** curt has joined #gnuenterprise
*** bigbrother has joined #gnuenterprise
*** bigbrother has quit IRC
*** curt has quit IRC
*** curt has joined #gnuenterprise
*** bigbrother has joined #gnuenterprise
*** SachaZzz has quit IRC
*** lupo__ has quit IRC
*** psu has joined #gnuenterprise
<psu> for the logs
<psu> (which, for once, means "reinhard, read this", not "psu, read this")
<psu> ;-)
<psu> yes, my bad re the creating objects at run time
<psu> I guess I just saw the word 'object' and assumied you meant AppServer object
<psu> not an object on a Form.
* psu blames the computer industry for lack of imagination
<psu> the word 'object' must have more contexts than just about any other
<psu> or alternatively
* psu blames the English languge
<psu> yeah, that'll do...
<psu> Anyway, will correct next time. Thanks.
*** psu has left #gnuenterprise
*** SachaS has joined #gnuenterprise
*** johannesV has joined #gnuenterprise
<johannesV> good morning
<SachaS> morning
*** sjc has joined #gnuenterprise
*** reinhard has joined #gnuenterprise
<reinhard> good morning everybody
<johannesV> good morning
<SachaS> hallo reinhard
<SachaS> reinhard: was there a lug the other day?
<SachaS> reinhard: I just saw you leaving but then I could not find any information on the lug vorarberg page about any meeting.
*** btami has joined #gnuenterprise
<reinhard> was an internal meeting regarding LinuxDay organisation
<reinhard> but I already proposed we should make these meetings public
<reinhard> next time
<SachaS> ok. thanks.
*** btami has quit IRC
*** dimas has quit IRC
*** ajmitch has quit IRC
*** btami has joined #gnuenterprise
*** dimas has joined #gnuenterprise
*** ajmitch has joined #gnuenterprise
<SachaS> btw wednesday a member of the Liechtenstein Landtag (a Landtagsabgerordneter) placed a small question (kleine Anfrage) for the Liechtenstein government (Regierung) about "Open Standards as well as Free and Open Source Software for the Liechtenstein Governement"
<SachaS> this Friday (later today) the government will have to answer the few questions asked
<reinhard> interesting
<SachaS> October 3rd a swiss study will be released about Free Software in Swiss Governement
<SachaS> infos: http://www.swissict.ch/oss.html
<SachaS> results of the Swiss Open Source Study
*** sjc has quit IRC
*** dimas has quit IRC
*** kilo has joined #gnuenterprise
<johannesV> bha, pyqt sucks
*** btami has quit IRC
*** kilo has quit IRC
*** lekma has joined #gnuenterprise
<lekma> hi guys
*** jamest has joined #gnuenterprise
<reinhard> hi lekma
<lekma> hi reinhard
<lekma> how are you?
*** johannesV_ has joined #gnuenterprise
<reinhard> fine, thanks
<reinhard> I just committed one of your patches yesterday
<reinhard> and wrote some comment about the other one in the tracker
<reinhard> (maybe you want to have a look at that)
*** johannesV has quit IRC
<lekma> i did see it
<reinhard> hmmm
<reinhard> why does it always have to be come friday 15:00 until I can start with the work I resolved to do this week...
<reinhard> s/be come/become/
*** johannesV_ has quit IRC
<lekma> why does it always have tobe come end of september until I can start with the work I resolved to do in april...
<lekma> :)
<reinhard> :)
<SachaS> why do i have to work in the first place?
<SachaS> why cant i just have fun?
<SachaS> why dont i have a ferrari in my garage?
<SachaS> why dont i own a villa in Mallorca?
<SachaS> why are there no gnuenterprise modules yet?
<SachaS> ;)
<lekma> reinhard: replied to your mail
<reinhard> thanks
<lekma> reinhard: do you have a little time to discuss?
<reinhard> sure
<lekma> while updating the ebuilds for gentoo
<lekma> i've been experimenting with appserver
<lekma> there is smthing wrong in the way it starts
<lekma> when you want it to daemonize
<lekma> btw: that's why the patch on the GDebug thing
<lekma> i use debug option for logfiles
<reinhard> yes, that makes a lot of sense
<reinhard> actually I'd even say
<reinhard> ajmitch: debian packages should also do something like start gnue-appserver with --debug-file=/var/log/... option
<reinhard> and maybe have some way to set debug levels in some config file
<lekma> ajmitch: i have an init.d file and a default one that could be adapted to debian, if you like
<lekma> reinhard: that's what i'm doing for gentoo
<lekma> anyway, while playing i stumbled upon a puzzling pb
<lekma> if you start appserver in daemon mode while giving it a debug-file option
<lekma> there a file descriptor open that wont be closed
<lekma> thus blocking the graceful closing of a session
<lekma> ie: ssh to your server
<lekma> start appserver with smthing like /usr/bin/python /usr/bin/gnue-appserver --silent --debug-level 0-2 --debug-file /var/log/gnue/appserver.log --pidfile /var/run/gnue/appserver.pid
<lekma> and try to exit your ssh session
<lekma> it hangs
<lekma> the only way to exit gracefully your ssh session is to stop appserver (kill -15 pid)
<lekma> sthe pb is that the GDebug happens before daemonizing
<lekma> (and daemonizing don't do the right thing, another pb (more a matter of taste in fact))
<reinhard> I'm trying to reproduce the "can't exit gracefully" thing now
<reinhard> what do you mean with "daemonizing don't do the right thing"?
<lekma> well, that's more a matter of taste, but according to http://www.erlenstar.demon.co.uk/unix/faq_2.html#SEC16
<lekma> you should close all open file descriptors
<lekma> and then reopen the to whatever you want
<lekma> smthing like:
<lekma>         try:
<lekma>             MAXFD = os.sysconf("SC_OPEN_MAX")
<lekma>         except (AttributeError, ValueError):
<lekma>             MAXFD = 1024
<lekma>         for fd in reversed(range(MAXFD)):
<lekma>             try:
<lekma>                 os.close(fd)
<lekma>             except OSError:
<lekma>                 pass # probably not open
<lekma>         os.open("/dev/null", os.O_RDONLY) # standard input (0)
<lekma>         os.open("/dev/null", os.O_RDWR) # standard output (1)
<lekma>         os.open("/dev/null", os.O_RDWR) # standard error (2)
<lekma> see, also: http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/278731
<bigbrother`> Title: Creating a daemon the Python way; Submitter: Chad J. Schroeder; Updated: 2005/10/03; Version: 1.5; Category: Threads; Description: The Python way to detach a process from the controlling terminal and run it in the background as a daemon.
<bigbrother> '\x02Title\x02: Creating a daemon the Python way; \x02Submitter\x02: Chad J. Schroeder; \x02Updated\x02: 2005/10/03; \x02Version\x02: 1.5; \x02Category\x02: Threads; \x02Description\x02: The Python way to detach a process from the controlling terminal and run it in the<br>\nbackground as a daemon.'
<lekma> :)\
<reinhard> okay
<reinhard> so this would also solve the problem with exiting ssh, right?
<lekma> yes, but it will create pbs with GDebug
<lekma> it's abit complicated
<lekma> the way GDebug works (if i understand correctly)
<lekma> is capture stderr and stdout
<lekma> and this is done before becoming a daemon
<lekma> so if you close all fds
<lekma> you lose GDebug (i tested it)
<reinhard> ok
<lekma> the only solution i see is forking in the background before anything that opens a file occurs
<lekma> do you think it is possible with the actual code?
<lekma> another question: is there someone working on a psycopg2 driver?
<lekma> if no, i'd like to give it a try
<reinhard> lekma: I think the daemonizing could be done at an earlier stage, before setDebug
<reinhard> FWIW, I can reproduce the problem
<reinhard> lekma: do you think you can dig into this?
<lekma> this, being?
<reinhard> i.e. doing the close/open stuff in daemonize, and moving the daemonize before the setDebug?
<lekma> unfortunately i tried already
<reinhard> in any case I suggest we create an issue in the tracker with some snippet of this log so the problem is documented
<reinhard> ok, I'll try
<reinhard> but I'm not sure when I will be able to do it
<reinhard> I'd like to have it in roundup
*** dimas__ has joined #gnuenterprise
<lekma> the pb is setDebug occurs in GBaseApp
<lekma> in __init__
<lekma> if you move daemonize before it
<lekma> you lose the distingo between a GBaseApp and a GServerApp
<lekma> daemonizing being being the main difference
<lekma> and if you move setDebug, you either have to do it in GBaseApp and GServerApp or ... you lose one of them
* lekma doesn't think he's perfectly clear
<reinhard> I know what you mean
<reinhard> I think the change would involve some restructuring of function calls
<lekma> that's why i couldn't provide a fix, it involves DECISIONS :)
<reinhard> lekma: I'm perfectly happy if you can open an issue in roundup with priority "bug"
<lekma> ok
<reinhard> and assign it to moe
<reinhard> err
<reinhard> me
<reinhard> I must run to the garden to pick some rasperries because sabine needs them urgently, bbl
<dimas__> moe on russian means - mine
*** dimas__ is now known as dimas
*** SachaS has quit IRC
<reinhard> back
<reinhard> lekma: thanks for putting the issue in roundup
<lekma> the issue with GDebug lead me to another question: shouldn't we use the logging module for logging purposes?
<reinhard> you mean syslog?
<lekma> nope, std logging python module
<reinhard> huh?
<reinhard> let me look
<lekma> i see 2/3 area where maybe using python standard module would simplify things
<reinhard> ah
<reinhard> logging module
<reinhard> new in python 2.3
<reinhard> probably nobody ever looked into it
<lekma> optparse for command line args
<lekma> ConfigParser for config file parsing
<lekma> sorry this one is already used
<lekma> and logging for... logging
<lekma> it was not included in python dist before 2.3, but it existed
*** SachaS has joined #gnuenterprise
*** derek has joined #gnuenterprise
<reinhard> logging module certainly looks like something we should switch to
<reinhard> I'm all for using built-in python stuff instead of homegrown extra stuff
<reinhard> I created an issue in roundup for this one
<reinhard> lekma: thank you for all your excellent proposals!
<lekma> you're welcome
*** jcater has joined #gnuenterprise
<lekma> is there any test suite i should run, if i want to test a new db driver?
*** psu has joined #gnuenterprise
<lekma> i created a roundup issue for a psycopg2 driver, with a patch
<lekma> (i can't believe that's the only work needed to add a db driver, kuddos to you guys!!)
<reinhard> gnue-common/test/drivers contains something
<reinhard> you might also want to add some info for technote 00016 in common
<reinhard> (I think the test given above gives you that info)
<lekma> well, the test fails for psycopg and psycopg2
<lekma> DB000: Traceback (most recent call last):
<lekma> DB000:    File "checkem.py", line 101, in ?
<lekma> DB000:     BackendCheck ().run ()
<lekma> DB000:    File "checkem.py", line 37, in __init__
<lekma> DB000:     GClientApp.__init__ (self, connections, 'common', {})
<lekma> DB000:    File "/usr/lib/gnue/common/apps/GBaseApp.py", line 345, in __init__
<lekma> DB000:     self.configDefaults, configFilename = self.CONFIGFILE)
<lekma> DB000:    File "/usr/lib/gnue/common/apps/GConfig.py", line 55, in __init__
<lekma> DB000:     self.loadApplicationConfig(configFilename,homeConfigDir,section, defaults)
<lekma> DB000:    File "/usr/lib/gnue/common/apps/GConfig.py", line 85, in loadApplicationConfig
<lekma> DB000:     parser = GConfigParser(defaults + GCConfig.ConfigOptions)
<lekma> DB000:  TypeError: unsupported operand type(s) for +: 'dict' and 'tuple'
<lekma> same error
<reinhard> oh
<reinhard> should be fixed now
*** psu has quit IRC
<lekma> yep, works
<lekma> python checkem.py > psycopg.txt
<lekma>  python checkem.py > psycopg2.txt
<lekma> diff -urN psycopg.txt psycopg2.txt
<lekma> <nada>
<reinhard> oh, nice
<lekma> (of course i changed psycopg to psycopg2 between run)
<reinhard> I just noticed that psycopg module is hosed in dapper
<reinhard> we need to document bugs in dependencies that hit us in our tracker
<reinhard> and document which versions of the dependencies they apply to
<reinhard> bah
<reinhard> no single postgres driver in dapper works :(
<reinhard> psycopg, pygresql and pypgsql all broken
<lekma> import psycopg2.psycopg1 as psycopg
<lekma> :)
<lekma> http://initd.org/tracker/psycopg/wiki/Migration
<reinhard> lekma: we wouldn't need that, AFAICT we don't need any of those features that changed between 1 and 2
<reinhard> off for today
<reinhard> cu all
*** reinhard has quit IRC
<lekma> bye all
*** lekma has quit IRC
*** dimas has quit IRC
*** psu has joined #gnuenterprise
*** psu has left #gnuenterprise
*** derek has quit IRC
*** derek has joined #gnuenterprise
*** dimas has joined #gnuenterprise
*** psu has joined #gnuenterprise
*** psu has joined #gnuenterprise
*** psu has left #gnuenterprise
*** jamest has left #gnuenterprise
*** jcater has quit IRC
*** derek has quit IRC
*** gambolputty has joined #gnuenterprise
<gambolputty> is there an IRC channel for Bayonne?
*** gambolputty has left #gnuenterprise
*** dimas has quit IRC
*** derek has joined #gnuenterprise