world as value

[Rolf-Thomas Happe]

I wrote:
>    Do you think that a model of the *World would make complex Clean programs
> more intelligible?  The merit of a Clean semantics for the I/O functions
> is not that obvious to me.  (Maybe it's just lack of education.  The question

I guess I should rephrase.
   Would an explicit denotational semantics of the I/O functions help not
only meta-software but also human programmers?  This is not obvious to me.  
   [Denotational semantics (which I don't really know, anyway) forces a
specification to be explicit.  Consider for example the extra lengths the
Scheme report has to go _not_ to specify evaluation order.  As for the
I/O functions, their denotations would answer more questions than I expect
I will ever ask.]

rthappe