mixed type operations in Octave

[Jaroslav Hajek]

On Mon, Sep 8, 2008 at 10:14 AM,  <soren at hauberg.org> wrote:
> Quoting Jaroslav Hajek <highegg at gmail.com>:
>>
>> 7. pretend that there always exists a real type wide enough to hold
>> all 64-bit integers, and use the semantics "convert both operands to
>> wide real, do OP, convert to intXX". On systems where this type really
>> exists (like the 80-bit long double on x86) it could be implemented
>> directly; otherwise, it would need software emulation (painfully slow
>> and laborious).
>
> Just out of curiosity, but would it be possible to have a long_double type
> in Octave, and would such a type solve the difficulties? Having a
> long_double type might be nice to have anyway.
>
> Søren
>
>

Well the main problem with this is that only basic arithmetic
operation would work on these (no BLAS/LAPACK).
It could still be benefitial, though. But what to do on architectures
where 80-bit extended prec is not present? Emulate? What if
quad-precision reals are available instead? Anyway, David's single
precision implementation showed us that adding a new type to Octave is
far from smooth.

So, should I take it that you vote for 7.?


-- 
RNDr. Jaroslav Hajek
computing expert
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz