design advice on predicate name

design advice on predicate name

[Marco Maggi]

Ciao,

  I have two predicate functions that act upon a vector
or matrix: the hit predicate produces a single #t/#f
result while the map predicate produces a vector or rank
2 array of #t/#f values.

  Using generic functions I can write functions with
math operators that act upon common numbers or vector and
matrices without visible difference.

  Example: I have two predicates HIT-NAN? and MAP-NAN?
which one it is better to call NAN?

  By the way: the NUMBER? predicate is supposed to return
#t for proper number SMOBs or for any object that can
be used as argument of math operators?


--
Marco Maggi

"They say jump!, you say how high?"
Rage Against the Machine - "Bullet in the Head"



_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: design advice on predicate name

[Neil Jerram]

"Marco Maggi" <marco.maggi-ipsu@poste.it> writes:

> Ciao,
>
>   I have two predicate functions that act upon a vector
> or matrix: the hit predicate produces a single #t/#f
> result while the map predicate produces a vector or rank
> 2 array of #t/#f values.
>
>   Using generic functions I can write functions with
> math operators that act upon common numbers or vector and
> matrices without visible difference.
>
>   Example: I have two predicates HIT-NAN? and MAP-NAN?
> which one it is better to call NAN?

I'm afraid I don't understand.  Perhaps you could write the down for
the two possibilities that you have in mind.

>   By the way: the NUMBER? predicate is supposed to return
> #t for proper number SMOBs or for any object that can
> be used as argument of math operators?

No, number? is supposed to return #t only for Guile's built-in numeric
types.

Regards,
     Neil



_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: design advice on predicate name

[Marco Maggi]

"Neil Jerram" wrote:
>"Marco Maggi" <marco.maggi-ipsu@poste.it> writes:
>>   Example: I have two predicates HIT-NAN? and MAP-NAN?
>> which one it is better to call NAN?
>
>I'm afraid I don't understand.  Perhaps you could write
>the down for the two possibilities that you have in mind.

For a vector of real numbers like this [1 +nan.0 3]:

* hit-nan? returns #t because at least one element
  is nan;

* map-nan? returns #(#f #t #f), one boolean for each
  element;

the same for matrices. map-nan? works like the 'isnan'
function of GNU Octave, which for the example vector
would return [0 1 0].

  hit-nan? can be used in the conditional of IF and
COND, while map-nan? must be inspected. For this reason
I guess that hit-nan? should be the nan?, but, to the
best of my knowledge, GNU Octave defines only the map-nan?
equivalent so I do not know how useful can be hit-nan?
in practice.



--
Marco Maggi

"They say jump!, you say how high?"
Rage Against the Machine - "Bullet in the Head"



_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: design advice on predicate name

[Neil Jerram]

"Marco Maggi" <marco.maggi-ipsu@poste.it> writes:

> "Neil Jerram" wrote:
>>"Marco Maggi" <marco.maggi-ipsu@poste.it> writes:
>>>   Example: I have two predicates HIT-NAN? and MAP-NAN?
>>> which one it is better to call NAN?
>>
>>I'm afraid I don't understand.  Perhaps you could write
>>the down for the two possibilities that you have in mind.
>
> For a vector of real numbers like this [1 +nan.0 3]:
>
> * hit-nan? returns #t because at least one element
>   is nan;
>
> * map-nan? returns #(#f #t #f), one boolean for each
>   element;
>
> the same for matrices. map-nan? works like the 'isnan'
> function of GNU Octave, which for the example vector
> would return [0 1 0].
>
>   hit-nan? can be used in the conditional of IF and
> COND, while map-nan? must be inspected. For this reason
> I guess that hit-nan? should be the nan?, but, to the
> best of my knowledge, GNU Octave defines only the map-nan?
> equivalent so I do not know how useful can be hit-nan?
> in practice.

Thanks for explaining.  My intuition is that the NAN? name should be
used only for a procedure which takes a single number and reports
whether it is NAN or not (as #t/#f).

For your `hit-nan?' procedure I would use the name `contains-nan?' or
`any-nan?'.  But on the other hand, if `hit-nan?' is the name that
first occurred to you, why not stick with it?  For your `map-nan?'
procedure I would use `map-nan?', as you have done.

Is it a feature of Octave that any predicate can automatically map
over a vector, or can `isnan' only be applied to a vector?

Regards,
        Neil



_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: design advice on predicate name

[Marco Maggi]

"Neil Jerram" wrote:
>My intuition is that the NAN? name should be used only
>for a procedure which takes a single number and reports
>whether it is NAN or not (as #t/#f).

Yes. The abstract behind having a NAN? for my custom
vectors and matrices is to let the user write math
functions that work for both numbers and composite
numeric objects. To let him try them out. I am not sure
about how significant it is, though.

"Neil Jerram" wrote:
>Is it a feature of Octave that any predicate can
>automatically map over a vector, or can `isnan' only
>be applied to a vector?

Predicates that act upon numbers can be applied to
numbers, vectors and matrices. And they always work
like MAP-NAN? Using the number/vector/matrix oriented
functions of Octave it is possible to write custom
predicates that behave in the same way.

Octave also has 'all()' and 'any()' which return a single
0/1 result if all or any of the elements of a num/vec/mat
is 0 or 1. So Octave requires two functions call to
achieve the result of HIT-NAN?

--
Marco Maggi

"They say jump!, you say how high?"
Rage Against the Machine - "Bullet in the Head"



_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user