From: SourceForge.net <no...@so...> - 2011-10-31 20:01:45
|
Bugs item #3422487, was opened at 2011-10-12 20:20 Message generated for change (Comment added) made by saggson You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=544942&aid=3422487&group_id=75752 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: Server Group: Latest Subversion Status: Open Resolution: None Priority: 5 Private: No Submitted By: saggson (saggson) Assigned to: Nobody/Anonymous (nobody) Summary: Fleets do no longer fight Monsters Initial Comment: I play FreeOrion r4349 atm. From turn #718 on Monsters are no longer defeated by my fleets. Each turn there are numerous messages like "'A battle occured at [some system]"'. But not one message about defeating any monster. At later turns they begin to visit the formerly monster-free inner systems of my empire and block supply lines. I have attached a save game for turn 716, with the 717 after that still showing a monster defeated message (and "psychogenic domination researched"), I'll keep save games 710-739 if needed. BTW There was not a single crash yet - great work! ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-31 21:01 Message: I just recompiled FO and replayed a few turns before and after researching psychogenic domination. - There were messages like "ERROR has been researched" - There were no crashes - There were no messages about battles - There were no messages about destroyed monsters - More and more monsters were taken over and 0wned by me -- not only in my systems but also during flight and in neighbor systems - messages about *monster* ship repairs were shown one turn later Anything else I could try? I'd really like to start a fresh game to have a more meaningful test. BTW gigi r1049 introduced a hard dependency on libboost 1.47 which cmake didn't warn about. Before that I could still get away with 1.46. I could still compile FO against my installed gigi snapshot. Could you make cmake warn about this? ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-26 04:25 Message: Could you try again to research Psych. Dom. with the latest SVN (code and data)? ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-20 22:53 Message: I just retested the same configuration but with an older save game that had no psychogenic domination queued yet. I could play for >20 turns without crashes. Then I queued and later researched that tech. Upon ending the turn after that - server crashed. ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-20 22:27 Message: Next test with r4422 and default/ at r4391, same scenario as before: Server crash as before after ~5 turns, server log attached. ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-20 21:41 Message: I have now redone the test of last night (as I described it) - reverting default/ to r4391 (the rest still r4417) and loading the turn before research on psychogenic domination completed. Again I emptied the research queue and ended several turns without enqueuing new tech research. The server crashed after five or six turns. I have "shortened" the server log to the last 112k lines or so -- starting with the time stamp of the last client log message "(HumanClientFSM) WaitingForTurnData.TurnProgress". That is still 14MByte of textual pleasure -- before compression. Now I'll do as you proposed most recently and retest with trunk HEAD. Should I keep the default/ folder at r4391 or also update it to HEAD? ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-20 07:17 Message: Also, I just committed some changes to lessen the chances of crashes when loading games with inconsistent tech names as the current techs.txt. If you could try again with everything at the lastest SVN, that would be interesting / useful. ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-20 01:01 Message: Can you post the logs for the crash after reverting default to 4391 and then emptying the research queue? ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-20 00:57 Message: Yes, psychogenic domination was the only one. I just tried reverting default/ to r4391 which didn't help. I loaded the turn before research on psychogenic domination succeeded, emptied the research queue and ended the turn and the next one (where the crash had occurred) without enqueuing any tech. Crashed again :-( Tried to load an older save game, where psychogenic domination had not been in the queue yet - game hanged... So there is probably some other problem. Do you have any other suggestions? Otherwise I'm going to start a completely new game this evening. If I am not quicker than last week that will probably take at least a few days. I'll report what I find out. ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-19 23:42 Message: Is the psychic domination tech the only one you research that turn? You appear to be having a crash while processing tech effects. I'd like to know what happens if you play a turn without researching anything, and with only researching psychic domination. There have also been a bunch of changes to the game data files that removed various techs and ship parts, and this is generating lot of noise in some of your log files (about null techs on queues). You might want to load a game before the psych domination tech researches, remove all techs from the tech queue, and re-enqueue anything you want (and hope that the tech screen actually loads with potentially missing techs. Alternative, you might want to revert the contents of your default directory to 4391 (approximately) for this save game. Or, I suppose you could check out the default directory separately for 4391 and set that to the content directory in the options screen (and then restart the game). ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-19 21:18 Message: I just attached logs of a turn change after researching "psychogenic domination". The other seems to miss a bit of relevant info. ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-19 20:43 Message: Uhm never mind the ascii art... ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-19 20:32 Message: My test with r4417 crashed the server upon starting the turn after researching "psychogenic domination". A save game of that turn is produced nevertheless. Loading it crashes the server immediately. I attached (nice and short) logs of such a loading attempt right after starting the FO client. freeoriond.log contains impressive ascii art :-) Let me know if that is not sufficient and you need logs of the crash after start of turn. ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-19 04:26 Message: I just committed a change that modifies how the SetOwner effect works. This should hopefully fix issues where ships have their owner changed, but their fleets do not. This won't fix the issues with existing saves that have this situation, but it might prevent it from happening again when that tech gets researched. ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-18 22:47 Message: There are numerous battles happening where empire 1 is the only empire with any ships present. Given that this starts happening after a tech that changes the owner of ships, I suspect what's happening is that the ships are changing ownership, but their fleets are not. This means there are fleets in the systems where battles are happening that are unowned (ie. monsters), and the combat system is deciding this should fight against the player's ships in that system. But when actually doing the combat, there are no non-player-owned ships in the system, so no battle damage or destruction of ships can occur. ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-18 21:17 Message: The log files for one example turn are now attached. According to sitrep battles took place in eight systems and in one system -- Euterpe -- two floaters were destroyed. But there are other systems -- at least Vargas and Burke -- where nothing has changed for many turns. Regarding the part about other empires: that was a reply to your first question about those - which you withdrew shortly afterward. Exactly before I sent my reply :-) I rephrased it later only for clarity. I don't think they were relevant to the monster battles at all. And regarding r4307 again: I never played that revision. The first FO build I ever tried -- r4349 -- failed in Universe.cpp I found the failing line changed in r4307 and reverse-patched the diff between r4306 and r4307 into my working copy (That's what the svn diff ... pipe patch -R does). This workaround is no longer necessary and I have already reverted it. ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-18 21:16 Message: Ahem to be more precise: Vargas and Burke have had monsters present for many turns now and every turn had "A battle occurred at [...]" but never messages about destruction of a monster. My fleets and the monsters also didn't loose health points during that time ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-18 21:11 Message: The log files for one example turn are now attached. According to sitrep battles took place in eight systems and in one system -- Euterpe -- two floaters were destroyed. But there are other systems -- at least Vargas and Burke -- where nothing has changed for many turns. Regarding the part about other empires: that was a reply to your first question about those - which you withdrew shortly afterward. Exactly before I sent my reply :-) I rephrased it later only for clarity. I don't think they were relevant to the monster battles at all. And regarding r4307 again: I never played that revision. The first FO build I ever tried -- r4349 -- failed in Universe.cpp I found the failing line changed in r4307 and reverse-patched the diff between r4306 and r4307 into my working copy (That's what the svn diff ... pipe patch -R does). This workaround is no longer necessary and I have already reverted it. ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-18 09:05 Message: Can you load a game when this is happening, and play a turn, then attach the server and client log files (zipped) to a comment, or post them elsewhere? There should be some logs related to combats in them. What is the relevance of the SVN diff / patch command you used? I ask about why you used 4307, not how you reverted to it... But if you're using the latest SVN now, it doesn't matter. Are your comments about other empires related to the endless monster battles? If you have a separate issue with other empires, make a separate bug report about it. If it's just that they don't do enough to oppose you, that's a missing feature, not a bug. ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-17 20:49 Message: To rephrase what I wrote about other empires: I have contact to one since the beginning of the game, destroyed some of their colony ships and have detected fleets of another but no attacks on planets took place yet. ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-17 20:31 Message: e.g. one floater with 10 hp against my fleet with - 6 Mark I as auto-designed by the game - 3 twin-Laser equipped in endomorphic hull (with defense grid/optical detection) - 4 twin-ion-cannon equipped in endomorphic hull (with deflector shields/sensor detection) - 1 twin-mass-driver equipped in endomorphic hull (with deflector shields/sensor detection) No one got hurt. On either side. I could send more ships. Any suggestions? Or am I completely overpowered? Regarding your earlier question about reverting r4307, I did this: svn diff -r 4306:4307 | patch -p0 -R ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-17 07:24 Message: Specifically what ships and monsters are involved in these battles? Could you send a more powerful ship to actually destroy the monsters? ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-16 23:38 Message: I have now built and tested FO svn revision 4402. No problems at all with building from source anymore. Misfortunately the battle problem persisted. Other than with r4349 some battles WERE resolved even after finishing "psychogenic domination". But a few turns later it's almost like I already reported: Many "A battle occured at [some system]" messages but very seldom (other than never as with r4349) about defeating a monster or even destruction of one of my ships. Turn after turn monsters and my fleets sit there at my systems having "a battle" occuring but not loosing any health points. Regarding your other questions: The(oretically) monsters were always able to move anywhere in my empire. They visited very frequently until my empire was strong enough. Now my fleets are in their way and normally defeat them before they move on. Sometimes they pass one perimeter system but I normally caught them soon afterward. I don't understand see what role other empires could play here. I have contact to one since the beginning for the game and have detected fleets of another but no attacks on planets took place until now. Any other tests I could do? ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-16 22:48 Message: Scratch the explanation in my previous post re: "If you psychically dominated another empire's planet...". The tech actually dominates ships, not planets. So, you might have taken control of a monster or enemy ship, leading to those battles being reported. So: where are those battles? Check that there's actually something wrong, and not that those battles should be happening and are being correctly reported... ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-16 22:45 Message: Did you need to build 4307 specifically, or was that just the first older version you tried? What newer version(s) didn't work? How didn't they work? If there are still issues building the latest SVN, please post about them. ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-16 22:38 Message: The issue with the save file is likely due to non-portability of binary boost-serialization, possibly related to you building on 64 bit Linux and me building as a 32-bit Windows binary. The tech in question might be working as intended. If you psychically dominated another empire's planet, and they had ships there, battles might ensue. You could also be fighting monsters at those planets, or fighting a battle between two planets each turn. Is researching this tech necessary for monsters to move into your previously-clear empire regions? It's not clear that this is a related problem... ---------------------------------------------------------------------- Comment By: saggson (saggson) Date: 2011-10-16 22:02 Message: I play on Debian testing amd64 and got the FreeOrion trunk sources with subversion, same with GiGi (r1055) and bullet (r2441). Everything else is from Debian packages. I redownloaded my save game to test if it has been damaged - still works. I tried to continue from there but stopped researching that "psychogenic domination" mentioned earlier. Two turns later I got reports about two victories against monsters. So that tech seems to trigger the problem... I will still svn update and retest but wanted to share this clue with you now. (BTW I had to revert r4307 to get freeorion built) ---------------------------------------------------------------------- Comment By: Geoff Topping (geoffthemedio) Date: 2011-10-15 19:54 Message: What OS are you playing on, and how did you get revision 4349? I can't seem to load your save file, so I can't test your issue, but there were some problems with the distance checking code fixed by revision 4352 and 4357 that might be related. Testing an up-to-date version would be useful. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=544942&aid=3422487&group_id=75752 |