Menu
▾
▴
[Colossus-tracker] [ colossus-Bugs-3604000 ] Battle Hangs When Trying to Carry Attack (Safari w/ Java 7)
|
From: SourceForge.net <no...@so...> - 2013-02-13 20:24:57
|
Bugs item #3604000, was opened at 2013-02-10 03:09 Message generated for change (Comment added) made by cleka You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=101939&aid=3604000&group_id=1939 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: Java Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Michael () Assigned to: Clemens Katzer (cleka) Summary: Battle Hangs When Trying to Carry Attack (Safari w/ Java 7) Initial Comment: Battle hangs when carrying over damage. The controls (e.g. concede battle) appear active but have no effect. Was working correctly before i upgraded to java 7. ---------------------------------------------------------------------- >Comment By: Clemens Katzer (cleka) Date: 2013-02-13 12:24 Message: I could reproduce that problem (on XP). 1.6.0_38 ok but _39 gets stuck. The problem was caused by disposePickCarryDialog. Not only was it not called inside EDT (but it should); it's also called from several places what is probably not needed. Trunk revision 5300 contains the "make sure it's called in EDT". Could some others try whether that resolves their problems too ? ---------------------------------------------------------------------- Comment By: Clemens Katzer (cleka) Date: 2013-02-12 07:03 Message: Regarding Davids report, sounds to me like the server side died/jumped out of it's main listening loop. So client sends something but server side does not process it any more. One thing coming to mind: the passing back of user's choice for such dialogs I have changes for some of them, some not. SOme of them might now be like: set a variable, close window, main thread proceeds; some others, directly write the result to the socket, then close window and continue. If there's something like that, it might in some ways make the message to be sent twice. And such behavior could be occur now due to how exaclty EDT handles stuff. Could also be whatever else, of course - Just speculating :) ---------------------------------------------------------------------- Comment By: Clemens Katzer (cleka) Date: 2013-02-12 06:58 Message: > although the game has started chewing > 300-500 MB of RAM after the Java update. Java 7 has (IIRC) a different default heap size (-mx argument). How do you start Colossus? Provided jnlp file and batch and script run it with -mx256. Normally the default of 64 MB is enough, except when you use the ExperimentalAI - then one might run into out of memory errors rather frequently with 64 MB only. Hence we changes it to be starting with 256 now always. Plus the actual JVM and stuff you easily get to 300-400. Current build - you mean the current official release 0.13.2. Yes, I know about those socket error; it has to do with a "internal dummy client" which I added to keep track of all messages that would have to be sent to each later added client - such as spectators ; on the game server, that is. On local games the client is created but the server accept listener expects only the "real amount of players". Newest sources from subversion should not show that problem. If I've time I will check it out to be sure. Thanks anyway! ---------------------------------------------------------------------- Comment By: David Ripton (dripton) Date: 2013-02-11 16:39 Message: I just tried it (Mac, 1.6.0_37), and on my first game, a battle hung after one of my creatures struck, presumably because it needed to ask me if I wanted to carry. The carry dialog never appeared, but the battle would not progress, as if the dialog were there and modal. The masterboard continued to work. Trying to close the battlemap window gave the "concede yes/no" prompt, but if I said yes it would not concede. Unfortunately I was playing from JWS and didn't open the Java console, so if there was a traceback or error log, I didn't see it. ---------------------------------------------------------------------- Comment By: bombasticus (bombasticus) Date: 2013-02-11 15:49 Message: Thanks -- this is the current build. I redownloaded to make sure. No Citrix or anything fancy. Plenty of RAM, although the game has started chewing 300-500 MB of RAM after the Java update. I'll report back on the "weird" non-carry hangs as they happen. One thing: every start generates two errors, a Socket Initialization Failed and a dummy An Error Occurred. They're non-fatal, so I've just been clicking through. ---------------------------------------------------------------------- Comment By: Clemens Katzer (cleka) Date: 2013-02-11 02:36 Message: And I suppose both of you are using the latest official released build, i.e. 0.13.2 ? The public test build is known to have some "hanging" bug related to conceding. (That's why there's still no new release yet). Bombasticus. thanks for contributing here instead of creating a new one! -Cle. ---------------------------------------------------------------------- Comment By: Clemens Katzer (cleka) Date: 2013-02-11 02:32 Message: Looks like it's the event dispatcher thread (the one which "listens" for mouse clicks and keyboard, and does the drawing) ran into a deadlock. There's another bug report which claims it works with 1.6.0_36 but hangs with 1.6.0_37, and yet another from Mac with 1.7.0_13 which says it hangs and there would be messages on stdout, but I suspect the messages are unrelated (seem to be common known issue with certain Java versions on certain MacOS versions. Question: Is there something special on that machine, like Citrix installed or so? There was a bug while ago where due to citrix a 2nd (security related EDTrelated thread) was running and those two interfered (caused each other deadlock). Can you go back to 1.6.0 so that you can play at least? (install 1.6.0 additionally and make Colossus run using that java) ? ---------------------------------------------------------------------- Comment By: bombasticus (bombasticus) Date: 2013-02-10 18:02 Message: Hi there -- definitely the carry. Doesn't happen all the time and the battles are hanging at other points (end of some battles, when clearing a dead creature other times) but this is probably the top source of my post-upgrade hangs. The good news is that the masterboard seems fine. Thanks as always for supporting this. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=101939&aid=3604000&group_id=1939 |