🚀 go-pugleaf

In article <DM4L7A.Esw@cix.compulink.co.uk> David Collier,
dcollier@cix.compulink.co.uk writes:
>re JPI
>
>They think that LONGCARD( 3.0 ) is 3. This is sensible, but not, perhaps, 
>in the spirit of purity.

Ok, I never even thought of doing that since also I had used more
standard M2 compilers before. This is how it must be then regarding 
"A := Type(B)" in pseudo M2 (sort of):

IF SIZE(A) = SIZE(B) THEN
   IF (A and B both are ordinal types or SETs) THEN
   	  bitpattern is reinterpreted (* PIM *)
   ELSE
      A := VAL(Type,B) (* useful but nonstandard *)
   END
ELSIF SIZE(A) > SIZE(B) THEN
   some bytes in A become undefined (* nonstandard and VERY dangerous *)
ELSE
   some bytes are lost  (* nonstandard, may be useful but still dangerous
                           - especially when porting code *)
END;

It is clearly safest to avoid using this extension (contamination?) and to
stick with the PIM case as I suggested and use VAL for the rest.


Regards,

.......................................................................
Bengt-Inge Dalenback
at:                                     and:
   Department of Applied Acoustics         CATT
   Chalmers University of Technology       (Software for room acoustic 
   Gothenburg, SWEDEN                      prediction and auralization)
   e-mail: bid@ta.chalmers.se              e-mail: catt@netg.se