[Haskell-cafe] No Derived Read for Unboxed Arrays

[Neil Mitchell]

Hi,

> I was trying to convert some code from ordinary boxed array to unboxed
> arrays for performance reasons.

> However my code ultimately failed because
> I load a large array saved as a text file using the derived Read, Show
> mechanism.

I found that Read was maybe 30 times slower than the slowest binary
serialisation method I could possibly think of. If performance matters
to you, and the array is more than a few elements long, switching away
from Read/Show should be the first step - before going for unboxed
arrays.

(But of course, having Read/Show defined for UArray may well be
useful, and sounds a good idea)

Thanks

Neil