Hi,

'c' < 'b' doesn't work in APL either but 'c'='d' works in J and in APL.



On Fri, Mar 1, 2019 at 1:15 PM 'Pascal Jasmin' via Programming <
programm...@jsoftware.com> wrote:

>  related,
> 'c' < 'b'
> |domain error
>
>     On Friday, March 1, 2019, 1:03:38 p.m. EST, Raul Miller <
> rauldmil...@gmail.com> wrote:
>
>  It does seem ironic that < is both "less than" and "box" but that
> "less than" on boxed arguments is a domain error.
>
> Perhaps, early on, that domain error was seen as more likely to be
> useful than a problem?
>
> (Most likely, though, doing tolerant comparisons on boxed arguments
> introduced high complexity issues for an operation like "less than",
> and our people who might have the insight to resolve those issues have
> had other things to do...)
>
> Thanks,
>
> --
> Raul
>
> On Fri, Mar 1, 2019 at 12:45 PM R.E. Boss <r.e.b...@outlook.com> wrote:
> >
> > For comparisons we have equal (=) and not equal (~:) and less than (<)
> and larger than (>).
> > I ignore min (<.) and max (>.) and less or equal (<:) and larger or
> equal (>:) since I consider them as direct derivatives of < and >. Also
> match (-:) is ignored.
> >
> > All four (=, ~:, < and >) have dyadic rank 0, but I am more interested
> in their differences, which, to me, are rather annoying.
> > We get
> >    =/2 4 ?.@$ 2
> > 0 0 1 1
> > for obvious reasons, and likewise with the 3 other verbs.
> >
> > But why is the next behavior different?
> >    =/;/2 4 ?.@$ 2
> > 0
> >    ~:/;/2 4 ?.@$ 2
> > 1
> >    </;/2 4 ?.@$ 2
> > |domain error
> > |      </;/2 4?.@$2
> >    >/;/2 4 ?.@$ 2
> > |domain error
> > |      >/;/2 4?.@$2
> >
> > Because we can compare the boxed items, as is shown by
> >    /: ;/2 4 ?.@$ 2
> > 0 1
> >
> > Needless to say also the other comparisons go wrong
> >    <./;/2 4 ?.@$ 2
> > |domain error
> > |      <./;/2 4?.@$2
> >    <:/;/2 4 ?.@$ 2
> > |domain error
> > |      <:/;/2 4?.@$2
> >
> > Of course I can circumvent the errors in one way or the other, e.g. by
> using /:, but usually that costs quite some performance.
> > Is there any time soon that the techniques which are used to compare
> boxed atoms, like used in /: , will be used for < and > too?
> >
> >
> > R.E Boss
> >
> > ----------------------------------------------------------------------
> > For information about J forums see http://www.jsoftware.com/forums.htm
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to