Re: [Bug-apl] Is the index generator function definition allowed to work on vectors of any length?

[Elias Mårtenson]

It's an extension. Since it's undefined by both APL2 as well as ISO 13751, it's a compatible extension, and a quite useful one at that. For example, creating a multiplication table is: ×/¨⍳10 10, which also extends naturally to more dimensions, unlike the alternative which uses ∘.×.

There was a discussion about this on this mailing list last year or so when the extension was added. My understanding was that the extension was taken from Dyalog, but you say it's not there, it must have been taking from somewhere else. Perhaps it was from NARS2000?

Regards,

Elias

On 11 April 2017 at 09:55, Leslie S Satenstein <address@hidden> wrote:

Re Gnu APL. 1,7-1

Gnu apl processes     ⍳ 2    5
as

 1 1  1 2  1 3  1 4  1 5

 2 1  2 2  2 3  2 4  2 5

and

⍳ 2    5    2

as

 1 1 1  1 1 2
 1 2 1  1 2 2
 1 3 1  1 3 2
 1 4 1  1 4 2
 1 5 1  1 5 2

 2 1 1  2 1 2
 2 2 1  2 2 2
 2 3 1  2 3 2
 2 4 1  2 4 2
 2 5 1  2 5 2

Since when does interval (index generator) take more than a scaler, or a vector with rho greater than 1 ?

Where would this (above example) be used?

Its not described in APL2, or the iso document.  I am searching dyalog's documentation and it's not there either.

Is this an interpreter error or some undocumented extension?

 

Regards

 Leslie

Leslie Satenstein

Montréal Québec, Canada