[Aldor-l] is sal_sfloat.as buggy?

[David Casperson]

Christian Aistleitner, on 2007-09-25, you wrote:

> Subject: Re: [Aldor-l] is sal_sfloat.as buggy?
> 
> Hello,
>
>> [ various variants of importing powf ]
>
> to me, it is not so important, how many or which functions work for now.
> It is more important to me, to hit the correct semantics of the code.
>
> According to the manpage of powf on my machine, powf is a function taking
> two C floats and returning a C float.
>
> Therefore, to meet the functions requirements, one has to import it in
> exactly that way, by using C float's equivalent Aldor type, which
> according to the AUG (Chapter 19.3; page 195) is SFlo.
>
> Regardless of which of the various import statements work or do not work,
> regardless of which of the various import statements produce the correct
> width or incorrect width for their identifiers, only the import statement
> using SFlo meets the specification of the AUG and powf:
>
>  import { powf: (SFlo, SFlo) -> SFlo } from Foreign C;
>
> It might well be that other formulations work as well (like using Rep
> instead of SFlo) but these formulations do not model powf adequately.

I think that it is an important question whether or not Rep can
be used in place of SFlo$Machine, I also tend to agree
with Christian Aistleitner that the SFlo formulation of the
declaration of powf is better than the Rep one, although I
believe that the semantics of Aldor say that they ought to be the
same.  My reason for the superiority of the SFlo version is that
if someone changes the Rep == ... line, and forgets to change the
definition of ^$SingleFloat, then the code ought to produce an
error message in the Aldor compiler itself rather than in the C compiler
invoked on its behalf.  (It seems that the latter aren't being
consistently noticed.)

Stephen, Oleg:  surely we can get the Aldor compiler to produce
the semantic equivalent of a reinterpret cast here (add a temporary,
and cast its address through struct { unsigned char[sizeof
FiWord] ; } *, or something equally ugly but formally correct).
Am I understanding the technical question correctly to be "How do
we translate "pretend" into C?

Cheers,
David
-- 
Dr. David Casperson, Assistant Professor     |  casper at unbc.ca
Department of Computer Science               |  (250)   960-6672 Fax 960-5544
College of Science and Management            |  3333 University Way
University of Northern British Columbia      |  Prince George, BC   V2N 4Z9
                                              |  CANADA