> Doesn't that remove the incentive to find an efficient all Haskell
> solution or find
> out why that is impossible (and maybe draw a conclusion from
> that, or is
> that just what
> is happening?).
Sure - that's why I'm reluctant to do it. There's another side to the
argument, though: if all we want is an array of bytes, why do we need to
force it through a Haskell compiler and have it type-checked etc.
Happy doesn't use arrays for its generated parsers by default, mainly
because Haskell arrays have all sorts of bounds checking and laziness which
means they are too inefficient for the simple "array of bytes" we need, and
GHC doesn't have any extensions to support static arrays (well, except for
compiling the array with the C compiler and using indexIntOffAddr# or
whatever).
Simon
