#pentagram@irc.freenode.net logs for 12 Nov 2004 (GMT)

Archive Today Yesterday Tomorrow
Pentagram homepage


[00:35:44] --> watt has joined #pentagram
[00:36:17] --- ChanServ gives channel operator status to watt
[00:53:11] <-- Kirben has left IRC ("System Meltdown")
[00:59:09] --> Kirben has joined #pentagram
[00:59:09] --- ChanServ gives channel operator status to Kirben
[02:40:08] --- Jett is now known as Darke
[05:40:29] --> Ember has joined #pentagram
[05:40:29] <-- Darke has left IRC (Read error: 104 (Connection reset by peer))
[06:27:14] --> Jett has joined #pentagram
[06:41:03] --> sbx has joined #pentagram
[06:45:09] <-- Ember has left IRC (Read error: 110 (Connection timed out))
[08:02:50] --> Colourless has joined #Pentagram
[08:03:00] --- ChanServ gives channel operator status to Colourless
[09:35:40] --- sbx is now known as sbx|afk
[10:40:40] <-- Kirben has left IRC (Read error: 110 (Connection timed out))
[12:24:05] --- Jett is now known as Darke
[13:09:23] <-- wjp has left IRC ("brb")
[13:09:45] <-- exultbot has left IRC (shutting down)
[13:09:59] --> exultbot has joined #pentagram
[13:09:59] --- Topic for #pentagram is: Spoon!
[13:09:59] --- Topic for #pentagram set by Darke at Sun Feb 15 13:15:28 2004
[13:11:17] --> wjp has joined #pentagram
[13:11:17] --- ChanServ gives channel operator status to wjp
[13:38:58] --> Fl00der has joined #pentagram
[14:00:33] --> WattAtWork has joined #pentagram
[14:19:45] <wjp> hi watt
[14:19:51] <WattAtWork> hi
[17:44:05] --- sbx|afk is now known as sbx
[18:22:51] <-- Darke has left IRC (Read error: 104 (Connection reset by peer))
[18:33:38] --> Darke has joined #pentagram
[20:11:00] --> edbgon has joined #pentagram
[20:12:20] <edbgon> hello all
[20:14:58] <wjp> hi
[20:18:29] <edbgon> how goes it?
[20:19:20] <wjp> still stuck on that cascading failure stuff
[20:19:41] <sbx> hi edbgon
[20:20:51] <edbgon> hey sbx
[20:23:10] <edbgon> that is too bad, i was amazed at how well it was working
[20:24:36] <edbgon> im sure it will soon anyway :-D
[20:25:13] <wjp> the longer I stare at it the more I think failures shouldn't cascade by default
[20:25:38] <wjp> which would fix this sequence and break the titans' death animations
[20:27:56] <edbgon> what causes the initial failure?
[20:28:58] <wjp> usecode explicitly killing a process
[20:29:07] <edbgon> ah ok
[20:33:08] <Colourless> shouldn't automatically cascade
[20:33:28] <Colourless> but i could be wrong
[20:33:34] <Colourless> case to consider
[20:33:44] <wjp> I'm fairly sure about that now as well
[20:33:47] <Colourless> what happens when you try to get the return value from a terminated process
[20:33:57] <wjp> but the titan death case really does have to cascade, as far as I can tell
[20:34:25] <wjp> you don't get the return value from a terminated process; the terminated process gives it to you
[20:34:55] <wjp> return value is an argument to the wakeUp call
[20:35:01] <Colourless> but if it never sets it, the value is garbage
[20:35:16] <wjp> true
[20:35:33] <wjp> of course if it continues cascading the garbage value will never be used anyway :-)
[20:35:41] <Colourless> yes
[20:36:14] <Colourless> a scenarios
[20:36:26] <Colourless> process A spawns process B
[20:36:33] <Colourless> process A waits for process B
[20:36:40] <Colourless> process B is 'terminated'
[20:36:58] <Colourless> process A continues and attempts to get return value
[20:37:04] <Colourless> value is 'invalid'
[20:37:18] <Colourless> do we teminate process A?
[20:38:10] <wjp> hm, that is an argument for cacading failures
[20:38:35] <wjp> (note that a termination is not a failure)
[20:38:58] <wjp> failures are caused by resetRef, objects being deleted and actors dying at this point
[20:39:31] <Colourless> that is what i meant
[20:45:34] <edbgon> terminate them all, mua ha ha
[20:46:37] * wjp hmmms
[20:46:59] <wjp> one other option is that the process that is being resetRef-ed now should already have terminated earlier
[20:47:17] <wjp> that would mean a sync problem
[20:47:32] <wjp> wait a sec...
[20:48:14] <wjp> I think this may answer the 'CHECKME' in ActorAnimProcess.cpp, line 232
[20:50:02] <wjp> if that fixes it I've just wasted a _lot_ of time :-)
[20:51:00] <edbgon> oh no :)
[20:51:38] <wjp> well, I was a bit 'out of' some usecode internals, so getting back into it was useful anyway :-)
[20:53:14] <edbgon> hehe
[20:54:45] <wjp> well...
[20:54:49] <wjp> fixed :-)
[20:55:24] <edbgon> really?
[20:55:38] <wjp> yes
[20:56:06] * Colourless examines this CHECKME
[20:56:42] <edbgon> hehe, that makes for good fun
[20:57:13] <Colourless> ok... now how did you know 'that' was the problem?
[20:57:20] <wjp> http://www.math.leidenuniv.nl/~wpalenst/edbgon-mordea if you want to continue playing
[20:57:51] <wjp> Colourless: I suddenly remembered that after saying that "one other option" line :-)
[20:58:06] <edbgon> well thank you very much
[20:58:19] <edbgon> will cvs be updated?
[20:58:31] <wjp> after I clean up locally :-)
[20:58:45] <edbgon> ok
[20:59:10] <wjp> things tend to be messy after long debugging sessions :-)
[20:59:41] <Colourless> as good as debuggers are, nothing beats debugging code :-)
[21:00:48] <edbgon> :)
[21:00:55] <edbgon> hmm, leiden.. i nearly chose to study there
[21:01:02] <edbgon> ended up in sweden instead
[21:03:09] <wjp> what did you study?
[21:04:01] <edbgon> well i was going to take the math requirement there... i study electrical engineering in the usa
[21:15:37] <wjp> ok, committed that
[21:15:45] <wjp> thanks for the bug report :-)
[21:17:00] <edbgon> heh, thanks for fixing it
[21:28:46] <-- sbx has left IRC ("brb")
[21:41:21] --> sbx has joined #pentagram
[22:07:20] <-- Colourless has left IRC ("casts improved invisibility")
[22:08:40] <-- Fl00der has left IRC ()
[22:13:51] <-- edbgon has left IRC ("Leaving")
[22:58:54] --> Kirben has joined #pentagram
[22:58:54] --- ChanServ gives channel operator status to Kirben
[23:20:18] <-- Darke has left IRC (Read error: 104 (Connection reset by peer))
[23:34:35] <-- WattAtWork has left IRC ("Chatzilla 0.9.66d [Mozilla rv:1.7.5/20041107]")
[23:35:39] --> Darke has joined #pentagram