🚀 go-pugleaf

In article <BxLxqu.96C@mentor.cc.purdue.edu> hrubin@pop.stat.purdue.edu (Herman Rubin) writes:
>In article <1992Nov12.082643.8415@arb-phys.uni-dortmund.de> wb@arb-phys.uni-dortmund.de (Wilhelm B. Kloke) writes:
>>Add Ada, Algol68, TeX as well.
>
>Some of these allow some flexibility in expression syntax.  But how
>many of these could conveniently unpack a floating point number into
>its exponent and mantissa?  In particular, could you write
>
>	exponent, mantissa =.U float
If this is the point, there is some clarification needed. You want to
have a result which is really *2* values. Most languages don't allow
this. You need a (possibly implicitly constructed) structure to make
this sort of assignment possible. Defining an unary operator which
returns a struct is possible at least in C++ and Algol68. As we know
Herman is interested in *very* efficient solutions, one could trie C++
with embedded asm code (conditionalised for the machines on which it can
be used) in the operator definition. If GNU C++ is this operator may be
even onlined. The syntax has to be changed to something like
struct {int exponent; double mantissa} e_m;
e_m Isplittedfrom floatvalue;
if the operator is called Isplittedfrom.
-- 
Dipl.-Math. Wilhelm Bernhard Kloke,
Institut fuer Arbeitsphysiologie an der Universitaet Dortmund
Ardeystrasse 67, D-4600 Dortmund 1, Tel. 0231-1084-257