--- Log opened Fri Sep 28 00:00:05 2012
09:02 < conseo> mcallan: the meinungsfindugstool guys use a pipermail mailing list as alex has posted: https://service.piratenpartei.de/pipermail/ag-meinungsfindungstool/
11:32 < conseo> mcallan: to close the db, we have to add a close (something called before destruction) routine to the VoteServer because the Database object is (imo correctly so) bound to VoteServer
11:39 < conseo> i think this is ugly. better to close the connection in each *Table class in a finally clause, so we can stop processing at any time. see this code: https://svn.apache.org/viewvc/commons/proper/dbcp/trunk/doc/PoolingDataSourceExample.java?view=markup
11:40 < conseo> i meant to close the connection in each table class method (immediatly) or if necessary document that result set, its statement and its connection need to be closed if the ResultSet leaves the class
11:48 < conseo> we don
12:13 < conseo> we'll forget what i said, it is not documented and i haven't found clear docs on the web either, but i am pretty sure ObjectPool has to be close()ed
12:14 < conseo> i miss destructors :-/ i guess this is what the Spool is for, but how to catch any termination of the program and close the db properly?
14:11 < conseo> i guess best is to use http://docs.oracle.com/javase/7/docs/api/java/lang/Runtime.html#addShutdownHook%28java.lang.Thread%29
14:49 < conseo> mcallan: u scared of parallel programming: http://www.kickstarter.com/projects/adapteva/parallella-a-supercomputer-for-everyone :-)
15:08 < mcallan> hmmm, i couldn't tell whether it was a joke or not
15:11 < conseo> which one?
15:13 < mcallan> (parallella) alex is correct, ofc.  shutdown hooks are available as a last resort.  but the immediate problem can be solved by adding to the Database "implements Closeable", and doing the cleanup there
15:14 < mcallan> whoever constructs Database is then responsible for calling db.close()
15:15 < mcallan> and since it's my own code that constructs Database, i can take it from there
15:16 < conseo> ok
15:16 < mcallan> re meinungsfindugstool.  i meant an english list
15:16 < mcallan> (or one with english component, as ed says)
15:16 < conseo> i will add a KickLogger interface and pass that to an UpdateKick on construction or otherwise use a default dummy one
15:17 < conseo> yes, i just saw that we can use the harvesting there already
15:17 < mcallan> ah, ok
15:23 < conseo> maybe KickTracker is better than "Logger", because it also calls a hook once the process is finished and might allow to expose other state changes like progress ?
15:30 < mcallan> i thought you were providing an interface for harvest to report its progress, not anything to do with kicks
16:04 < conseo> ok, i can do so, i cannot track progress for an update then though, it will just print out whatever it wants
16:07 < mcallan> i was only making a suggestion about the name
16:08 < mcallan> if it reports harvest progress, then maybe you want to call it "harvest something"
16:13 < conseo> well somehow we want the harvester to talk back to the callee. setting that in the kicker (which also makes it a single instance) is somewhat unnecessary imo
16:14 < conseo> if i would add a callback to the kick, the harvester could talk directly to the callee and expose information for each individual update
16:17 < conseo> maybe this is a bad idea, but i won't make the handler more complicated, it is just the question where to pass it to the kicker
16:17 < conseo> harvester i mean (damn i need to check-read more often sry)
16:20 < mcallan> (i was thinking of plan we discussed yesterday.  did not know you changed it.  no problem)
16:21 < conseo> i am just asking. atm. the kick only contains to final strings and is pretty much final. i ask whether adding a handler on construciton makes the architecture worse
16:26 < mcallan> i can turn on skype, but can only talk for 5.  must fix bugs
16:27 < conseo> ok
16:27 < conseo> ring me
16:49 < mcallan> (it's a tough one, and affects talk track.  will explain when i have a better grasp of it)
18:39 < conseo> oh, ok.
21:14 < conseo> mcallan: when i try debug the stage on Votespace with votorola/b/gwt-devmode, i always get: Cross-site hosted mode not yet implemented.
21:15 < conseo> what am i doing wrong or on which page do you debug it?
21:20 < mcallan> Wicket votespace, right?
21:21 < mcallan> you need to do this: votorola/b/build xgwt --module=votorola.s.gwt.stage.StageInDev
21:22 < mcallan> or if you will navigate to other (non-Wicket) pages with embedded stages, then you might want to build dev versions of all the modules
21:23 < mcallan> see gwt-devmode for the full command
21:25 < conseo> ok, thx, i see
21:51 < conseo> mcallan: i am off for today, gn8
21:53 < mcallan> n8 c
21:55 < conseo> voharvest behaves now as discussed btw., obeying the --verbose parameter
22:01 < mcallan> excellent, we're in better shape everyday :-)
--- Log closed Sat Sep 29 00:00:23 2012