As I mentioned, bsh is actually quite fast (a LOT faster than that - typical for me is 5 to 10 times slower), but there's a bug which if you trigger it somewhere in your script it will result in a massive massive slowdown (somewhere around 100-500 times IIRC). So, I would guess you're hitting that bug regularly

?
[...]
Well, loops are very slow...
for(int k=0;x<50;x++)if(foo[k]==16)foo[k]=0;
That's already a big drag...
for(int k=0;k<50;k++)for(int y=0;y<50;y++)if(foo[k][y]==16)foo[k][y]=0;
... and the framerate is down to... like 10fps with a 500mhz cpu. The same in java yields 500+fps and it's limited by the graphic card's fillrate.
It's horrible slow. bsh-2.0b2... maybe I should give b4 a try, but the changelog doesn't seem to mention any speed fixes. Ah well, I don't really mind it's slowness... execution speed isn't critical for prototyping
