matslina / bfoptimization Goto Github PK

The multiplication loop optimization appears to imprecisely preserve the behavior of the original Brainfuck code; in the mandelbrot program, with this optimization, the program will attempt to read/write out-of-bounds addresses on the tape. It works because malloc is used instead of mmap, thus, segfaults are avoided, probably due to over-allocation.
The same problem can be observed when trying to use the optimization in an interpreter.

I have been trying to figure out why this is the case for some time now but have yet to succeed.

EDIT: It appears that the fix is to disallow subtractions in multiloops? My mistake; not the problem.

EDIT 2: Looking back, it's obvious. Hindsight and such. The problem is that by optimizing multiloops, we were removing the conditional... which means behavior isn't preserved precisely. Simple fix is to just add an if statement before doing the multiplication operation.