Path: utzoo!utgpu!watserv1!watmath!att!rutgers!cs.utexas.edu!asuvax!ncar!ico!haddock!ima!esegue!compilers-sender
From: anw@maths.nott.ac.uk (Dr A. N. Walker)
Newsgroups: comp.compilers
Subject: Re: Enumerated data types
Message-ID: <1990Aug29.142641.28548@maths.nott.ac.uk>
Date: 29 Aug 90 14:26:41 GMT
References: <1990Aug23.134826.2865@forwiss.uni-passau.de> <3621@goanna.cs.rmit.oz.au> <2019@charon.cwi.nl>
Sender: compilers-sender@esegue.segue.boston.ma.us
Reply-To: anw@maths.nott.ac.uk (Dr A. N. Walker)
Organization: Maths Dept., Nott'm Univ., UK.
Lines: 25
Approved: compilers@esegue.segue.boston.ma.us

In article <2019@charon.cwi.nl> dik@cwi.nl (Dik T. Winter) writes:
[in Algol]
>	'proc'('real','int')'real' + = ('real' r, 'int' i)'real': ....
	^^^^^^ 'op'
> [...].  The only kind of coercions allowed for operands
>is 'dereferencing', [...]

	Also deproceduring and uniting, but not widening, rowing or voiding.
The reason is, of course, to prevent ambiguity -- you are not allowed to
overload operators with operands varying [eg] only in the reference levels,
but you are with operands varying [eg] only in row levels, so you can
disambiguate operands with the wrong reference levels, but not with the
wrong row levels.

	I suspect that at least rowing could have been moved into the other
camp [though there may be some subtleties involving the interaction between
rows and references], but pragmatically one may well want to overload, in
particular, "*" to mean different things for scalars, vectors and matrices.

-- 
Andy Walker, Maths Dept., Nott'm Univ., UK.
anw@maths.nott.ac.uk
-- 
Send compilers articles to compilers@esegue.segue.boston.ma.us
{ima | spdcc | world}!esegue.  Meta-mail to compilers-request@esegue.