[Haskell-cafe] Re: Haskell and C++ program

[Derek Elkins]

On Thu, 2009-01-15 at 13:40 +0100, Apfelmus, Heinrich wrote:
> Eugene Kirpichov wrote:
> > Well, your program is not equivalent to the C++ version, since it
> > doesn't bail on incorrect input.
> 
> Oops. That's because my assertion
> 
>    show . read = id
> 
> is wrong. We only have
> 
>    read . show  = id
>    show . read <= id  (in the "less defined than" sense)

No, you only have
read . show = id which often doesn't hold in practice.
show . read </= id

Assuming the first identity holds, you do of course have show . read .
show = show and this probably holds even in most cases where read . show
= id does not hold.