Re: [Bug-apl] value error when using 'Cut'

[Jay Foad]

FYI Dyalog are considering introducing Cut into a future version of
Dyalog APL and would welcome feedback or collaboration on the design:

http://video.dyalog.com/Dyalog15/?v=9KOto3xXS3c
http://www.dyalog.com/uploads/conference/dyalog15/presentations/D16_Future_Operator_Proposals.zip
http://www.dyalog.com/blog/2015/11/response-to-feedback-on-cut-under-and-merge/

Jay.

On 8 February 2016 at 01:58, Louis de Forcrand <address@hidden> wrote:
> AFAIK, Rationalised APL was the model for Sharp APL (which by the way is
> still downloadable for DOS from the
> resurrected Waterloo APL archives) and eventually led to J. The way Sharp
> APL deals with nested arrays is
> slightly different from ISO APL or APL2 (or any other APL2-like
> implementation), and uses < to “box” any array
> regardless of its shape (scalars too!) and raise its level by one (as does
> J), unlike enclosing does in the standard.
> In fact, many (quite useful) features such as composition and, by extension,
> inverses of composed functions, are
> either missing or not completely implemented in the ISO standard.
> I haven’t read Rationalised APL, but I wouldn't expect everything in there
> to work out-of-the-box in GNU or any
> other standard APL; I’ve tried SAPL and it is quite similar to J (and so
> quite different from the standard).
>
> Best regards!
> Louis
>
> On 07 Feb 2016, at 20:32, <address@hidden>
> <address@hidden> wrote:
>
> Jürgen,
>
> Thanks for clearing up the usage.
>
> I pulled my example from this paper:
> http://www.jsoftware.com/papers/RationalizedAPL.htm    (search the page for
> the term 'tessellation' to see where I am looking)
>  which seems to be using < in place of ⊂ ....which still seems to cause an
> error for me in GNU APL.
> The more I read that paper, it seems that it is intended as suggestions to
> add on to APL, rather than act as a reference...
>
> Is the tessellation available in GNU APL, and I am just messing something up
> (might be b/c I'm not sure what you mean by "literals")?
>
>
> -------- Original Message --------
> Subject: Re: [Bug-apl] value error when using 'Cut'
> From: Juergen Sauermann <address@hidden>
> Date: Sun, February 07, 2016 5:27 am
> To: address@hidden, address@hidden
>
> Hi Alex,
>
> not sure what   a ¯3⍤< m  is supposed to mean.
>
> According to the ISO standard the syntax for ⍤ is:
>
> Z ← f ⍤ j B           (monadic, page 124) or
> Z ← A f ⍤ j B         (dyadic, page 125)
>
> If you compare that with your example:
>
> a ¯3 ⍤ < m
>
> then the (expected value) j is the primitive function <, which triggers the
> VALUE ERROR.
> The fact that the caret points to a is not because a is the culprit, but
> because a is the left
> end of the phrase being reduced.
>
> Unfortunately the syntax in the ISO standard is somewhat ambiguous: j is a
> one, two, or three
> element vector, and B is the rest. Therefore it is sometimes impossible to
> decide where j ends
> and where B begins, and the examples for ⍤ in the ISO standard are in
> conflict with the IBM APL2
> binding rules. This  conflict occurs only with ⍤ which - wise decision - is
> not implemented at all
> in IBM APL2.
>
> The conflict can be avoided by always putting j and B into separate
> variables. If you use literals for j
> or B, heaven forbid, then be prepared for fairly nasty error messages at
> times.
>
> /// Jürgen
>
>
>
> On 02/06/2016 05:17 PM, address@hidden wrote:
>
>
> Hi bug-apl,
>
> Why am I getting a value error here? It seems that the variable 'a'
> definitely exists:
>
>       a←2 2 ⍴2
>       m←4 4 ⍴⍳16
>       a
> 2 2
> 2 2
>       m
>  1  2  3  4
>  5  6  7  8
>  9 10 11 12
> 13 14 15 16
>
>
>
> VALUE ERROR
>       a ¯3⍤<m
>       ^
>       a
> 2 2
> 2 2
>
>
>
>
> SVN 693
> -Alex
>
>
>