---Randy MacDonald wrote:
> Hello Ric;
>
> This thread seems to go around a J 'gotcha', the first I can recall
> which works this way, as opposed to things like #.^:_1 which
> work as you
> would expect them to, if you repressed your expectation of an error.
[snip]
> Indices suggested by u in u}y are allowed to be in terms of
> the ravelled
> noun y, but those suggested by m in m}y cannot be.
>
> Fortunately,
>
>    x m"_ }v
>
> is a valid workaround.  NB. My J epiphany of the day.

Good thinking. The workaround provides an easier (non-tacit) way to create u 
for u}y for the cost of a bit of leanness.

   10 ts '_&(([: I. 0 = ,@])}) M'
0.10898725955 71304640
   10 ts '_ ([: I. 0 = ,@])} M'
0.10935403103 71304512
   10 ts '_ (I. 0 = ,M)"_ }M'
0.11008181207 113247488
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to