Hi Jay,
I believe that when they wrote the ISO standard they were still
used to the old APL, where
f and g had to be scalar functions and then f/A g B of two vectors
would anyhow be a scalar,
and therefore they simply forgot to enclose the result of f
If you followed this part of the ISO standard verbatim then:
2 3 +.⍴ 4 5 ←→ +/2 3 ⍴ 4 5 ←→ 13 14
GNU APL and APL2 actually return ⊂13 14 which seems more sensible. I am not
sure why ISO has (or needs) that special case. It seems wrong to me.
Jay.
On 18 May 2018 at 16:03, Juergen Sauermann
