Brainfuck v2 implementations are broken

[kostya]

Brainfuck v2 implementations are broken

[kostya]

this is strange, i not see any bug in implementation.

[kostya]

i think some bug in bottles.bf file, i find another example: https://raw.githubusercontent.com/andrewtholt/4th/1574fc7e3ed1a3a4d6085fca19469c125ae3f390/4th/apps/bf/bottles.b, this works ok

[brk]

Using bf.py from the brainfuck2 directory:

$ python2 bf.py BOTTLES.BF
99 Bottles of beer on the wall
99 Bottles of beer
[...]
91 Bottles of beer on the wall
91 Bottles of beer
Take one down and pass it around
^C

Do you get different results?

The semantics of brainfuck are given on the Wikipedia page by translation to snippets of C code. In particular, [ and ] corresponds to a while-loop.

bf2c.hs does literally that: it generates snippets of C code for each character of brainfuck input.

$ runhaskell bf2c.hs < BOTTLES.BF > gen.c
$ gcc gen.c -o gen.exe
$ ./gen.exe
99 Bottles of beer on the wall
99 Bottles of beer
Take one down and pass it around
98 Bottles of beer on the wall
[...]
1 Bottle of beer on the wall
1 Bottle of beer
Take one down and pass it around
0 Bottles of beer on the wall

$

The linked BOTTLES.BF file produces different output in the two implementations. That means that either the Python interpreter has a subtle bug, or there's a bug in the translation into C code, right?