Note that (0 0$0), i. 2 2 would also work.  The trailing dimension
gets expanded, just like in your original case, so the net result is
equivalent to (0 2$0,i. 2 2 which is to say nothing happens.

But, depending on your situation, 0 0 $ 0 might be more generally
useful than 0 2 $ 0.

FYI,

-- 
Raul

p.s. sorry about the hijacking

On Sun, Jan 13, 2013 at 4:05 PM, Steven Taylor <[email protected]> wrote:
> I think the email thread got hijacked... but a follow up to the original
> join question...
>
> Roger pointed out that I wasn't being logical (i.e. appending rank 1 to
> rank 2 doesn't make sense).
>
> ... and if I do this, I now get back what I expected to see.
>
>     (0 2$0), i.2 2
> 0 1
> 2 3
>
> So... I borrowed the idea of q's "reshape" and implemented it in J as a
> helper verb.  So far, it seems to be doing the trick.
>
> thanks for the comments and help.
>
>
> On 13 January 2013 02:19, Raul Miller <[email protected]> wrote:
>
>> On Sat, Jan 12, 2013 at 8:50 PM, Linda Alvord <[email protected]>
>> wrote:
>> > As long as the ranks agree,  u&v y -> u v y
>> >
>> > Would this be useful as a definition of compose?
>>
>> That is the definition for the monadic case but is incomplete since it
>> does not mention the dyadic case.
>>
>> But note also that the dictionary's vocabulary is a reference work and
>> it has already stated that the ranks must agree.  Keep in mind that
>> the dictionary also includes material such as
>> http://www.jsoftware.com/help/dictionary/samp13.htm
>>
>> FYI,
>>
>> --
>> Raul
>> ----------------------------------------------------------------------
>> 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