Bug:Turn and view battle result not the same

Foodstamp · #1 January 6th, 2008, 03:38 PM

Sounds like your units could not clear the field before the turn limit.

lch · #2 January 6th, 2008, 07:25 PM

Okay, jokes aside, this discrepancy between battle replay and battle result is well known, probably caused by differences in operating systems of server and host, i.e. treatment of random numbers, and quite often is reported by people sporting an Apple Mac OS system. Check the search function for more info.

Agrajag · #3 January 7th, 2008, 07:37 AM

Quote:

lch said:
Okay, jokes aside, this discrepancy between battle replay and battle result is well known, probably caused by differences in operating systems of server and host, i.e. treatment of random numbers, and quite often is reported by people sporting an Apple Mac OS system. Check the search function for more info.

IIRC the specific bug you are referring to has been fixed.

thejeff · #4 January 7th, 2008, 10:01 AM

There is definitely a version of this bug in v3.10, that does not require different patch levels, operating systems, etc. I've seen it in 1 machine SP games, which I had not seen before 3.10

Agrajag · #5 January 7th, 2008, 10:52 AM

Quote:

thejeff said:
There is definitely a version of this bug in v3.10, that does not require different patch levels, operating systems, etc. I've seen it in 1 machine SP games, which I had not seen before 3.10

You are talking about several different things.
"Different patch levels" is one thing. A difference between patch versions or in the content of mod files between the hosting computer and the viewing computer causes a difference between the battle report and the viewable simulation because the host resolves the battle using certain unit stats that are different from the client viewing the simulation, and so you get different results. This is not really a bug, it's just a glitch you get because you used incompatible (well, at least "not fully compatible" versions of the game. (ie is your fault

)
"Different operating systems" is another thing. In this case the difference was caused because different checks in turn resolution employed something that looked like "if (random()+modifiers1>random()+modifiers2) then etc." and on some OSs the random() on the left was called first, while on others the right random() was called first, which basically "flips" the results you get in random rolls between the hosting machine and the viewing machine, leading to different combat replays than what actually happened. I believe this bug has been resolved.
So this is not "a version of the bug" but a completely different bug (currently unexplained) that has similar (but not identical) effects.
If it really is caused by v3.10, my guess would be that this patch introduced some change between the code used to resolve battles when hosting and the code used to resolve battles when replaying them.

EDIT: its => it's

thejeff · #6 January 7th, 2008, 12:12 PM

You are of course correct, but it seems largely a matter of semantics.

From one point of view a bug is the set of symptoms, since that's all you can see from the outside. A particular reported "bug" may result from more than one problem in the code. And a problem in the code may result in multiple apparently unrelated symptoms. Without finding a code problem and testing, you really can't tell.

I was speaking in terms of the symptoms, since that's what the users report and since patch levels and operating systems are always suggested as the source of the problem.

As for it being "completely different", it may be, but if this really is new in 3.10, the first place I'd look is in the fix to the old bug.

Torin · #7 January 7th, 2008, 12:46 PM

In the shortlist its reported like being a different os incompatibility but it happens even on SP play where theres only one OS and no server-client.

Agrajag · #8 January 7th, 2008, 01:40 PM

Quote:

thejeff said:
You are of course correct, but it seems largely a matter of semantics.

From one point of view a bug is the set of symptoms, since that's all you can see from the outside. A particular reported "bug" may result from more than one problem in the code. And a problem in the code may result in multiple apparently unrelated symptoms. Without finding a code problem and testing, you really can't tell.

I was speaking in terms of the symptoms, since that's what the users report and since patch levels and operating systems are always suggested as the source of the problem.

As for it being "completely different", it may be, but if this really is new in 3.10, the first place I'd look is in the fix to the old bug.

Yes, technically this is just a matter of semantics, I just argued it to make sure that it is indeed just a semantics issue, and not a case of a wrong understanding of things

(especially since you called a the difference in replay and actual events caused by different versions (of patches or mods) a bug, which it hardly is.)

Also, I meant that it is probably (but not necessarily) a completely different issue, I probably should have phrased it differently. As for looking in the fix to the old bug, that's pretty obvious since the fix is the code that directly controls the battle simulation - it's the first place you'd think to look regardless of previous bugs