slow ai (v1.17): source of problem possibly located

[Helmet]

I created a battle scenario in development version 1.17 between the player's 35 units and the ai's 20 or so units. Although the armies aren't that big, the ai was painfully slow during its turn. I'm not used to seeing the spinning beachball in Wesnoth, nor the hourglass. Wesnoth's ai has always been super-fast on my computer.

The ai would spend around 10 seconds to calculate each unit's move -- sometimes much longer than 10 seconds.

I think I discovered why, by accident. The culprit appears to be the new "gulch" tile.

When the ai was slow, the map had around a dozen gulch tiles with lava at the bottom. (They look awesome, by the way. I like the mist over the lava.)

Well, I recently replaced all the gulch+lava tiles with regular lava tiles and suddenly the ai is fast again!

I hope this observation was helpful toward stabilizing 1.17.

Thanks for working on this game.

[Helmet]

Sorry, I spoke too soon. The slow ai has returned. I guess the removal of the gulch+lava tiles was a coincidence. I have no idea why the ai was fast for a one single game and then slow again.

I'm on an M1 Mac, OS Ventura. I just checked the Activity Monitor. It said that The Battle for Wesnoth was taking 100% of the CPU.

[Pentarctagon]

Did you get Wesnoth from Steam or from elsewhere?

[Helmet]

I downloaded it from the website.

[Pentarctagon]

Huh, alright. And in the Activity Monitor under the Kind column for Wesnoth it says "Apple", right?

[Helmet]

Huh, alright. And in the Activity Monitor under the Kind column for Wesnoth it says "Apple", right?

Yes.

See attachment.

[Pentarctagon]

Can you provide an example save file where the AI is taking a long time?

[Helmet]

Can you provide an example save file where the AI is taking a long time?

Sure.

I'm play testing a new faction of snow golems in scenario 2. You might take a look in scenario 2 at the special powers used by certain snowgolem units to make sure that I didn't do something bad with a filter. But I didn't experience any slow-down when play testing scenario 1, and the units didn't change much from scenario 1 to scenario 2.

Good luck finding the problem.