No SPRITE commands are allowed in the Paint event, because they could cause an infinite loop.

Gill wrote on Jul 4^th, 2010 at 11:08pm:



Calling a sprite command (without the true parameter at the end) causes an immediate screen update, which causes the Paint Event to be fired.

Therefore, if you call TINTSPRITE (or any other sprite command) from within Paint, it will cause Paint to fire again, which will then call TINTSPRITE, which then causes Paint to fire, which then calls TINTSPRITE, etc...

seems sc crashes whenever theres a bug in the mouse move event

1) type "movesprite 1,0,0" in mouse move event
2) run game
3) move mouse
4) SC go byebye
5)IF forgot to save THEN 

EqwanoX wrote on Jul 10^th, 2010 at 2:16pm:



I've been trying to fix this but I can't seem to solve it.

What I do know - is that sometimes REALbasic generates two Mouse Move events in a row. The first Mouse Move event generates the scripting error. When you click Debug Game the entire game is shut down and the editor is brought back up, but the second Mouse Move event is still in the queue, and once RB gets around to it, the original GameWindow is gone so the program crashes.

Gill wrote on Jul 24^th, 2010 at 11:33pm:



It wouldn't matter - the only option for a scripting error in a compiled game is "Quit". Even if it quits unexpectedly (crashes), it doesn't matter. The real problem is running your game in the editor when you haven't saved your work. That's why you should enable "Save on Run" in the preferences.

In an earlier version (2.0 alpha) I had implemented a fix for this problem - instead of killing GameWindow, I hid it, and killed it after 1 second. This allowed any remaining events in the queue to complete successfully. Unfortunately this caused several other problems so I took out the fix. Probably though - short of a newer RB fixing this problem, it IS the only effective fix.

I still use REALbasic 2009r4 since newer versions may break things arbitrarily. I probably won't upgrade until forced to do so, OR if a newer version has substantial improvements. Honestly, the only reason I upgraded from REALbasic 5.5.5 was for Universal Binaries.

Al Staffieri, Jr. wrote on Jul 25^th, 2010 at 3:55pm:


I could probably do that. I already have a boolean called "running" that I set to false when the game has an error, and a bunch of checks everywhere (none of the events will actually execute a script if running is false). When there's a script error, I can set running to false, wait a second, then show the error box and kill the game. Any events in the queue that trigger another script to run, such as a timer, will see that running is false and won't run.