Fraser Jackson wrote:
> If you are wanting different sized samples I suggest you
> write SAMPLES so it works for unboxed arguments. Then ...

>From the description provided by Leigh, the case that
concerned him was where the arguments were the same
size, but the results were differing in size (the left
argument specified the size of the result.

A variation on Henry's suggestion would actually work
rather nicely for this case:

50 100 <@SAMPLE"0 2 y1

(Though I should add: if the definition of SAMPLE was
such that the left argument was always assumed to be
an atomic number and the right argument was always
assumed to have no more than 2 dimensions, I'd be
inclined to put "0 2 in the definition of SAMPLE.  The
only exception here would be if I were trying to debug
SAMPLE -- the debugger behaves nicer when explicit 
by themselves have names.  But I do not think that
debugger limitations should coding style outside of
the debugging context.)

-- 
Raul

----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to