Yeah, maybe. The amend applies to 'those parts of y selected by m&{'
which is nothing in each case; but 'an error is signalled if such
selection requires fill' which you could argue has happened when the
'selected parts' have shape 0 3 3. The problem with that argument is
that if fill was used on (0#a:) it was also used on (0#0).
My guess is that the amend has a hidden requirement that the selected
parts must have rank not exceeding that of y, which is a Good Thing
normally. But I don't find that in the Dictionary.
Henry Rich
Raul Miller wrote:
> On Sat, Jul 10, 2010 at 8:37 PM, Henry Rich <[email protected]> wrote:
>> $ (0#a:) { 3 3$'b'
>> 0 3 3
>> $ (0#0) { 3 3$'b'
>> 0 3
>>
>> That might account for the difference. I am not sure that the letter of
>> the Dic indicates that the amend should fail in the first case but not
>> the second.
>
> I would think that scalar extension should support both cases?
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
