>From experience with SML, I believe that any benefit from explicit
precedence definitions is far outweighed by the cost.  It complicates the
implementation (e.g. requiring more information to be shared across
compilation boundaries), it complicates the syntax (requiring new
declaration formats), and IMO it can actually make programs harder to read
(because you can't mentally parse infix expressions without checking the
precedence declarations).

Dave.


-----Original Message-----
From: Charles Martin [mailto:martin@chasm.org]
Sent: Wednesday, January 03, 2001 15:38
To: caml-list@inria.fr
Subject: Re: New Year's resolution suggestions...



>You mean that the actual implicit way of specifying associativity and
>precedence of users's defined operators is not powerful enough for your
>programs ?

It is probably powerful enough as it stands.  But it restricts me to the 
naming scheme that has been chosen for me, which I don't like.  What if 
some kind of Hungarian notation for alphanumeric identifiers was enforced 
by the compiler?  That would be awful.  This feels kind of the same.

This is not nearly as big a wish as for some kind of overloading! :)

Charles