[Haskell-cafe] Re: Aim Of Haskell

[Joachim Durchholz]

Claus Reinke schrieb:
> 
> but on the Pascal note: is there anything in Pascal that Haskell doesn't
> provide, and improves on (nested procedures, procedure parameters,
> distinguishing in and out parameters, types, ..)? it has been too long 
> since my Pascal days, I don't remember..

Nothing that I'm aware of. You'd have to be careful which version of 
Pascal you mean, there were lots of dialects around.

In general, however, I'm not sure whether contrasting Haskell to Pascal 
is a fruitful exercise. Pascal and C are nearer to each other than 
Haskell is to either of them after all. (Type classes, anonymous 
functions, type inference, just to name the first three that occurred to 
me...)

> apart from the communication problem of understanding Haskell as Pascal: 
> if you're talking to someone who knows Pascal, it might not be
> a bad idea to position Haskell as a drastically modernized version of
> Pascal, to get the discussion of real merits going?

No, not at all.
IMHO.

I think that Haskell is a step ahead of OO. The connection is a bit 
tenuous, but if you carry the Liskov Substitution Principle to its 
logical consequence, you end up disallowing any semantic changes in 
subclasses... and that means you don't need interface subclassing at all.
And to implement those inhomogenous lists and iterators and whatnot, you 
show how you can do that in a functional language without the 
subclassing baggage.

... it might be useful to show how the design patterns from the 
Gang-of-Four book can be done in a functional language. And with less 
restrictions.
Such a side-by-side comparison might help convince library writers and 
system architects (and these are among the more important people to win 
over anyway).

Regards,
Jo