--- Log opened Mon Apr 02 00:00:59 2012
06:32 < mcallan> conseo: the build message wasn't clear enough.  it wants you to install this version of gwt: http://zelea.com/project/gwt/release/
06:32 < mcallan> (fixed)
08:59 < conseo> will this be upstream?
09:22 < mcallan> not sure
09:41 < conseo> hmm, do you have a patch?
09:41 < conseo> i can fetch the tarball, but it will take some time
09:45 < mcallan> it's only 85M, you are on a slow connection?
09:45 < conseo> you can also post the patch on some gwt mailing list
09:46 < conseo> nope, well, do as you wish, i can fetch it and generate the patch myself
09:46 < mcallan> what do you need a patch for?
09:46 < conseo> at least i can see what is changed then
09:46 < conseo> curiousity
09:46 < conseo> and this should go upstream if possible imho
09:47 < conseo> but well just my 2 ¢
09:47 < mcallan> no reply yet https://groups.google.com/forum/?fromgroups#!topic/google-web-toolkit/U6qrVhP_9-o
09:49 < mcallan> patch is here: http://zelea.com/var/db/repo/gwt/rev/9ca0f5ccace5
09:50 < conseo> cool, thx
09:51 < conseo> you need to remove your copyright and assign that to google i guess, but they will tell you anyway
09:53 < mcallan> screw that, google can pay me $ if they want to become an anonymous drone ;-)
09:53 < mcallan> (want me)
09:53 < conseo> :-)
09:56 < conseo> ok, patching is difficult because of the jars they ship
09:59 < mcallan> (he doesn't trust me)
10:20 < conseo> trust is worse than understanding :-)
10:22 < conseo> mcallan: in HarvestWAP i create a voteserver and database connection on initialization (constructor), is this done for each request? is this ok?
10:22 < conseo> CountWAP looks like doing all internal setup and computation in the constructor and only parse the output in respond()
10:23 < conseo> so this happens for each connection (?)
10:23 < conseo> ah ok, you take the voteserver from req.wap()
10:23 < conseo> hmm
10:25 < mcallan> that's right
10:26 < mcallan> (it's very expensive to construct, so all WAP interfaces share a single one)
10:28 < conseo> ok, that is reasonable
10:30 < conseo> this wap interface is your demultiplexer for web access, right?
10:32 < mcallan> yes
10:53 < conseo> hmm, i get a lot of segv fault when building lately. often cleaning helps, but it is a bit weird
10:57 < mcallan> from our build scripts?
11:22 < conseo> i think javac has problems with stale class files (yes). but this is new, i barely had that with java 6
11:22 < conseo> might be related to some new code though
11:23 < conseo> (but it also happens inside rebuilding the harvest package, which does not do something special)
11:27 < mcallan> i get nothing like that.  maybe try switching to oracle's jdk, see what happens
11:32 < conseo> i don't want to use the oracle jdk. mind you that it works, i have never failed to build with clean before
11:33 < conseo> (i do so to debug though, but i will focus on writing that WAP stuff now)
11:53 < conseo> should i call the wap servlet hHarvest or hDiffCache ?
11:58 < mcallan> maybe better to keep it consistent with class name, so hHarvest i guess
12:06 < conseo> ok
12:06 < conseo> mcallan: what is going on in toronto: http://www.cbc.ca/news/canada/toronto/story/2012/03/31/toronto-occupy-arrests.html ?
12:10 < mcallan> not sure, first i heard
12:19 < conseo> ok
13:09 < conseo> mcallan: i have to add it to WAP.java:105 to make it work, right?
13:15 < mcallan> hmmm, i guess so.  not sure what i coded there, but that should work for you
13:16 < mcallan> (ah, ofc.  they need to be registered to support the short names)
13:17 < conseo> ok
--- Log closed Tue Apr 03 00:00:15 2012