tain.totalcodex.net

Everyday you see damage scores not adding up in myth, and its easy to see why with integer n floating point values. However, that taken into account, ive never quite seen this before. I dont see how 1:31 > 1:32 lol.
Prolly not a bug just never seen it before =D

Have no idea if this is related, but in June I posted a bug report/film whereby you can see one of the player's total koth flag time should be one value (they should win if no one contests - which they didn't), but slowly loses 3 to 4 seconds per minute as the end of the game approaches.

By the end of the game I end up winning (and no, it's not a special game-altering Baak plugin), even though it should have been the other player.

Was the first time I'd ever seen that. Usually adding up the player's time and the remaining clock time stays consistent.

It could have *nothing* to do with anything, but interesting to me that it might be the same kind of "loss of precision" problem. That's the kind of conincidence that gets my attention as a software engineer.

The original bug post is here: KotH Player Time Lags Clock

Baak wrote:The original bug post is here: KotH Player Time Lags Clock

fyi, that post no longer exists.

Doh! Sorry... it does still exist, but is in the "beta testers" forum and thus is not visible by all...

I've copied and pasted it here - thanks for catching that, Vinyl!


---vvv----------- KotH Clock Bug? Post -----------vvv---

Played a KotH (TFL) game tonight and noticed that a player's time on the flag lagged the game clock by about 3 to 4 seconds per minute in the closing 1:15 of the game.

I've posted the film here (zipped), and you'll need the Trebuchets plugin from the OoH Downloads page to watch the film.

It was a very close game - and very fun - but watch starting at about 1:15 remaining. At the point where Miltiades kills my last unit at the flag - from then on no one else touches or contests the flag. If you slow the game down and do shift-F7 to keep the times up (*love* that feature btw) you'll see that his time on the flag plus the time remaining starts off adding up to between 1:24 and 1:25, if he were to hold it and no one else were to contest it.

Milt pulls a great move and blocks my entrance to the center fortress (this being on Venice) - at around that point I added up the times during the game and was thinking it was going to be really close.

But watching the film it goes from Milt having a sure win with 1:24 to 1:25 (I have 1:22 total at that point, which is what I will end the game with), to Milt's total being 1:24, then 1:23 to 1:24, then 1:23, etc., right down to the end of the game where suddenly his total ends up at 1:21 (it's at the 1:20 to 1:21 cusp near the end).

Thus you can clearly see his time on the flag deteriorates from 1:24 to 1:25 down to 1:20 to 1:21 within a 1:15 timeframe. That's 3 to 4 seconds per minute.

Would suck to lose on something like that!

Anyway, never noticed this before - usually adding up times in-game is exactly the way they end up. Only went back to check because Milt seemed sure he had a lock when he took the flag - and it turns out he did!

Let me know if you have any problems getting the film or the Trebuchets plugin. Trebs on Venice is very fun.


Slight bit of potential quirkiness: Milt was the only one of the four of us running 1.5.1 (Two Saks dropped earlier in the game too). I'm sure testing KotH (TFL) (and maybe plain Myth II KotH?) should show if this happens consistently.

Hey Baak,

I found the bug.

Turns out Myth was eating two ticks of scoring every second because of a blah<0 instead of blah<=0 comparison.

For the record, you still win, with 1:28 to 1:27 (since your score was affected by the bug also). However, it wasn't all consistent. If you became eliminated at that point (with the buggy scoring), then Mil would have won, since the game time remaining is added up when game ends.

1.5.1 gameplay mode will still use the old scoring, and 1.6 will use the fixed version. Note: the fixed version still "appears" to be off by one second, but that is only because the last second (0) still counts as a full second, so you'd have to add 1 to the time remaining to get the result that Myth will compute in the fixed version.

As always, thanks for the bug report!

nice find myrd