MUM combat crash

[thunderchero]

Hi Everyone,

I have gotten this crash log on MUM lately, and I was able to reproduce crash every time so started to do some testing.

File: tcmath\tcmath.c, Line: 34, REAL_GE_ZERO(value)
Initialize State: 31

My current test saved game combat Klingon vs cardassians large combat 138 klingon heavy cruiser II vs 43 command, 111 fast attack, 32 strike 6 non combat.

I first encountered crash giving assault command vs fast attack after first 4 turns of combat 80% of the time.

since I expected evade command was cause, I started to give evade command to my heavy cruiser II's on second turn of combat (after giving assault once to get ships closer). after giving evade command second time (3rd turn) I get crash as soon as I press turn button 99% of the time.

after testing different agility values with only decreasing crash frequency I started looking at new patches that had been applied.

I did not expect any change, but when I removed "AI Command Probabilities Fix" patch I was unable to reproduce crash.
So next I applied only part of the patch (slow group 2) but this only delayed crash to turn 6

I really like this patch since it really improves AI commands in combat, but don't think crash chances is worth it.
I just don't understand why it is crashing.

[Spocks-cuddly-tribble]

File: tcmath\tcmath.c, Line: 34, REAL_GE_ZERO(value)

Main question is: LOD switch issue or random bug in the 3D combat calculations?

In fixed vanilla (LOD fix) and ECM the issue is still present but VERY rare IIRC.

ECM uses all AI commmands patches which might indicate a LOD issue in MUM?

when I removed "AI Command Probabilities Fix" patch I was unable to reproduce crash.
So next I applied only part of the patch (slow group 2) but this only delayed crash to turn 6

I really like this patch since it really improves AI commands in combat, but don't think crash chances is worth it.
I just don't understand why it is crashing.

'AI Command Probabilities' does not change any evade percentage (unless you did), but it (+ other AI TC command patches) will change the AI random commands, so this will affect tests but is not source of the issue.

Below AI patches should affect AI evade commands: viewtopic.php?p=52751#p52751

- Improving AI Combat Commands for Warships vs enemy Ships (UPDATED)
- AI Battle Modifier -> 'Escort Steadiness' low or off 0 -> try 100 (disable new fast group smart evade if shields down)
- Different Solo Commands within AI Combat Groups

Also MUM might still use an outdated AI evade (instead of retreat) patch? viewtopic.php?p=48957#p48957

EDIT:

I first encountered crash giving assault command vs fast attack after first 4 turns of combat 80% of the time.

since I expected evade command was cause, I started to give evade command to my heavy cruiser II's on second turn of combat (after giving assault once to get ships closer). after giving evade command second time (3rd turn) I get crash as soon as I press turn button 99% of the time.

Also note there is a bug with charge/assault vs evade -> ships can't move but only change orientation.

That's why I had to remove emergency evade vs cloaked fleets in ECM...

Above crash issue might be due to auto target switch after current target destroyed when using charge/assault. Old/new target evade vs none evade (possible with different solo commands). So movement calculations might get confused due to the bug. But this is untested speculation.

[thunderchero]

Main question is: LOD switch issue or random bug in the 3D combat calculations?

LOD switch would crash just as 3d combat started iirc (could be avoided if user would rotate screen sometimes)
This crash happens just as turn button is selected (this combat has 20 - 30 sec calculation time)

- AI Command Probabilities Fix patch - MUM uses same patch ecm uses
- Improving AI Combat Commands for Warships vs enemy Ships (UPDATED) - MUM was using outdated patch, but issue still present after update
- AI Battle Modifier -> 'Escort Steadiness' low or off 0 -> try 100 (disable new fast group smart evade if shields down) - MUM has edited patch but with ecm version or Escort Steadiness set to 0 or 100 issue still persists
- Different Solo Commands within AI Combat Groups - MUM uses same patch ecm uses

Also MUM might still use an outdated AI evade (instead of retreat) patch? viewtopic.php?p=48957#p48957

- outdated AI evade - patch not applied

this crash is just annoying

I did not expect any change, but when I removed "AI Command Probabilities Fix" patch I was unable to reproduce crash.

I have to correct this statement I would say 25+ combat attempts without "AI Command Probabilities Fix" patch 1 crash now with same log.

[Spocks-cuddly-tribble]

This crash happens just as turn button is selected

This is strange and annoying.

In another post you stated the crash might be related to agility most when using evade (re-cloak command shares the evade movement code).

GAL mod ships are very fast in combat, so this mod should be prone to this crash?

If not, I'd try to remove AI solo commands for testing. Since other crash was with UDM3 this only makes sense if UDM also uses AI command patches?

Improving AI Combat Commands for Warships vs enemy Ships (UPDATED) - MUM was using outdated patch, but issue still present after update

An outdated version had a wrong code part leading to crashes with no log, so I hope this was corrected in MUM (I deleted/updaded the patch a long time ago).

[thunderchero]

This is strange and annoying.

agreed

if UDM also uses AI command patches?

that report was from 2012 so I doubt it had many AI command patches at that time.

I'd try to remove AI solo commands for testing.

on limited testing (2 combats) no crashes without solo command patch, but same as removing AI Command Probabilities Fix it might just take many tests and still occur? btw this combat averages 20 combat turns.

[Spocks-cuddly-tribble]

I don't think you can eliminate this crash in massive fleet mods. It's about the best reasonable compromise to lower the frequency.

As told the 3D calculations charge/assault vs evade/cloak are bugged.

Removing AI solo commands weakens the AI but lowers ship spread out and distance within combat groups (if this is a factor for the crash?).

[thunderchero]

I don't think you can eliminate this crash in massive fleet mods. It's about the best reasonable compromise to lower the frequency.

As told the 3D calculations charge/assault vs evade/cloak are bugged.

Removing AI solo commands weakens the AI but lowers ship spread out and distance within combat groups (if this is a factor for the crash?).

something that I noticed when I wanted to view the tevent.txt the timing/command changed when -Mudd F11 fog of war active.

without F11
turn 1 group charge, turn 2 group evade, turn 3 no command given (ships still on group evade) .... crash

with F11
turn 1 group charge, turn 2 no command given (ships still on group charge), turn 3 no command given (ships still on group charge) .... crash

as for the tevent.txt the next turn is written. it has a new line with only "TURN: 3"

and testing is getting frustrating now lol

if anyone is interested here is the saved game (MUM current beta 3b)

game11.sav

(416.02 KiB) Downloaded 22 times

[Spocks-cuddly-tribble]

timing/command changed when -Mudd F11 fog of war active.

Yes, mudd cheat features seem to affect some in game random features (e.g. F11 changes known races for AI diplomacy features).

Could also be due to the BotF random engine fix included in ECM AI TC multi patch if installed.

[thunderchero]

Could also be due to the BotF random engine fix included in ECM AI TC multi patch if installed.

not applied to MUM

But I think I found the issue, when I replace the Ragnor model no more crash even with old outdated code.

The only thing I can think of is overlapping polygons of model is causing issue?

[Spocks-cuddly-tribble]

So basically sort of a variant of the 3D model LOD feature issue?

Btw if mod is ready/finished with beta tests, I'd add 'BotF random engine fix' (patch code auto skips MP games to prevent any issues).

[thunderchero]

So basically sort of a variant of the 3D model LOD feature issue?

new model was a fail still getting same error.

with new model crash timing changed with same commands
turn 1 group charge, turn 2 group evade, turn 3 no command given (ships still on group evade) repeat turn 3 after 8 turns .... crash

I did notice (limited testing) if I re-issued the combat orders each turn no crash. so far 3 test combats 19 turns each combat test.

[Spocks-cuddly-tribble]

I did notice (limited testing) if I re-issued the combat orders each turn no crash. so far 3 test combats 19 turns each combat test.

This sounds strange, but there is some issues with some combat orders.

E.g. HOLD has to be re-issued each turn (gets auto-changed by BotF code) and more important Charge/Assault have crash issues with wrong/missing target data: viewtopic.php?p=29642#p29642 plus the no movement issue vs Evade (messy code ).
So if all old targets (or all ships in target group) are destroyed it might trigger a similar issue, albeit I never noticed such a problem with vanilla and ECM.