ENIGMASH verbose_logging maxes out CPU

[angilbert]

Describe the bug
ENIGMASH verbose_logging prevents game play

Reproduction Steps

1. Load https://david-pfx.github.io/PuzzleScriptNext/src/editor.html
2. Load ENIGMASH example game
3. Start game and confirm movement is possible
4. Click on the button to enable verbose_logging, click back on the game
5. The game stops responding to input

What did you expect would happen?
verbose_logging works - as it does in it's original fork:
https://jacklance.github.io/PuzzleScript/editor.html?hack=cfdcc6e23f1fb3e9de2fd42fafaf4d4c

OS & Browser version, desktop/mobile
MAC/Desktop/Chrome (also tested Firefox)
Dev and Stable release behave the same

Depending on the browser you may eventually see a pop-up similar to this...

[david-pfx]

6. Click on the game.
7. The game now responds to input.

It's a kind of known bug, but the bug is about loss of focus, nothing more. I've known about it, but never got around to doing anything about it.

[angilbert]

That isn't the explanation here, but I retested it, to be sure. You can reproduce the issue a different way, by adding "verbose_logging" to the prelude first. Then save and run - and hit X to continue past the credit message. The screen then goes blank and the browser appears to hang.

After a long pause the game eventually appears.
In the browser console you can see an event handler takes 93 seconds to react:

[david-pfx]

I agree, I saw that too. The 'prevents game play' is incorrect, but horribly slow is a better description. My machine is a high end gaming laptop, and it appears to take 100% CPU on one core for nearly 30s. Enigmash is slow at the best of times, and verbose logging shows over 3000 rule applications.

My guess is one could put a lot of work into this and improve the speed by 20%. I don't think that's a priority, but getting the focus to work right probably is worth it.

[angilbert]

I've adjusted the title. Yes, I agree, that's a lot of rules. I did wonder if local_radius was broken, but comparing the verbose_logging from a single player move (step left) on the two forks side-by-side I see a comparable number of rules are triggered. That said PSNext does trigger a few more late rules, which is curious. Why should more rules trigger in PSNext? Both sources are 2580 lines - so I assume identical.

Here I am comparing...
https://jacklance.github.io/PuzzleScript/editor.html?hack=cfdcc6e23f1fb3e9de2fd42fafaf4d4c (left)
with https://david-pfx.github.io/PuzzleScriptNext/src/editor.html (right)

Question... Is local_radius actually working at all? On either fork?

[david-pfx]

That's weird. I'll have to look into what those extra rules are.