--- Log opened Tue Sep 04 00:00:11 2012
01:03 < conseo> mcallan: prototype state saving works
01:03 < conseo> good reason to go to bed
10:38 < mcallan> cool
14:39 < conseo> mcallan: i will write some test cases and then turn to the client side + the wap for kicks?
14:41 < conseo> i have separated the classes in the 4 subpackages cache,kick,run btw.
14:45 < conseo> mcallan: what do the points on the track mean? i remember they were related to vote circles, but it would make sense to add some information to the mouseover effect imo
14:46 < conseo> i don't understand it atm. without searching the list for the design discussion
14:47 < conseo> (cache, kick, run and auth)
15:17 < mcallan> i don't like the talk track i mocked up.  i wouldn't code it...
15:23 < mcallan> the only aspect of the design i like is the overall structure, a time series.  even there i think the direction is wrong.  the newest posts should be on the right side
15:25 < mcallan> conseo: so i would code only that part for now, if i were you.  a series of colourless dots (or something) growing from the left edge of the track
15:26 < mcallan> add the kick button (or whatever) to that.  code the rest only when you have a good design
15:27 < mcallan> (that would be my plan for the poll track, too)
15:35 < conseo> yes
15:36 < conseo> mcallan: i will separate the kick triggering from harvestwap and update from the client 10secs after i have sent a kick, or should i use some callback?
15:38 < mcallan> callback from server to client?
15:41 < mcallan> server cannot easily open connection to client.  nor can client easily hold a connection open.  i think you probably want to poll, yes
15:42 < mcallan> re points on track: http://zelea.com/project/votorola/s/gwt/stage/_/mock/home-1.xht
15:44 < mcallan> the quarter and full moons (which i would just make dots, for now) are diff messages
15:44 < conseo> mcallan: well, i could delay the callback from the server for 10 secs or so
15:44 < conseo> the javascript callback function from harvestwap
15:45 < mcallan> true...
15:46 < mcallan> i think it will hurt you.  you'll be holding too many connections open, could be thousands on a busy server
15:46 < conseo> yes
15:46 < mcallan> tomcat will be offended
15:47 < conseo> sure, it is not sane
15:47 < conseo> well, i will update then
15:47 < mcallan> sure, put the client to work
15:47 < conseo> yes
15:48 < conseo> this is the only way to slowly push the load to the clients. for now we can ignore scaling problems, but on the way we can push more and more to the client with html5
15:49 < mcallan> right
17:43 < conseo> mcallan: what was the event problem with gwt?
17:45 < mcallan> event problem?
23:20 < conseo> "gwt code is garbage in places.  they emulate java.util.HashMap as a list.  they do get by exhaustive search.  and the whole gwt event system is based on HashMap"
23:24 < conseo> i can't find it atm.
23:26 < mcallan> oh that.  i doc'd it here: http://zelea.com/project/votorola/_/javadoc/votorola/g/web/gwt/event/package-summary.html#package_description
--- Log closed Wed Sep 05 00:00:28 2012