Server-crashing / SP performance degrading bug in trigger.action.removeMark(), can be re-created at will

[cfrag]

Hey all,

    at long last I tracked down a severely performance-degrading (and in servers halting/crashing) bug: when a script invokes trigger.action.removeMark() on a mark that was drawn before the mission had a few cycles of mission drawing, the thread executing removeMark() crashes or slows down to a crawl. 

This can happen in any mission script that first draws a rectangle during init, and then immediately tries to erase and redraw it (for example if data was loaded at mission start) before the mission has really started.

The result is a mission that may run in single-player, slows down to a once every 10 second update in player-hosted MP, and completely trashes the server in dedicated.

I could submit a simple demo to recreate this bug, but ED appears to be so swamped with the other bugs that I've reported in the past years that I'll save me the trouble.

[Flappie]

1 hour ago, cfrag said:

I could submit a simple demo to recreate this bug, but ED appears to be so swamped with the other bugs that I've reported in the past years that I'll save me the trouble.

Performance-related bugs are usually fixed quickly when they are easily reproducible. Please post an example so I can report it.