Thanks Raul.

AtopA looks close enough to Brian's suggestion that it appeals to me.

On Thu, Jul 26, 2012 at 4:34 PM, Raul Miller <[email protected]> wrote:
> On Wed, Jul 25, 2012 at 11:47 PM, Ian Clark <[email protected]> wrote:
>>> m=: v^:x y                      NB. dyadic(&) --> (^:)
>>> n=: u"v m                       NB. Brian's suggestion for (@)
>>
>> Bearing in mind what Raul said earlier, I don't think this is general
>> enough. Maybe (@) is unavoidable in dealing with how u and v operate
>> together.
>
> It's not unavoidable.
>
> Here's an implementation of atop that does not use @
>
> atopA=: 2 :'([: u v)"v'
>
> Here's another approach:
>
> atopB=: 2 :'[: u@> <@v'
>
> That said, note that if side effects are involved, atopB reschedules
> those side effects, and this can be an issue (if the schedule was
> important).
>
> 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

Reply via email to