[Haskell-cafe] Proof that Haskell is RT

[Andrew Birkett]

Hi,

Is a formal proof that the Haskell language is referentially 
transparent?  Many people state "haskell is RT" without backing up that 
claim.  I know that, in practice, I can't write any counter-examples but 
that's a bit handy-wavy.  Is there a formal proof that, for all possible 
haskell programs, we can replace coreferent expressions without changing 
the meaning of a program?

(I was writing a blog post about the origins of the phrase 
'referentially transparent' and it made me think about this)

Andrew

--
- http://www.nobugs.org -