Leigh wrote:
> I've noticed that with a rank adverb (i.e., ,"_1 / ) your prefered
> solution works.
Raul countered:
> $,/"_1]i. 1 2 3
> 1 6
Ok then:
$ ,&]/ i. 1 2 3
2 3
$ ,&]/ 1$'a'
By definition, ,&]/ and ,/ can't differ; but since ,&]/ isn't
supported by special code, it doesn't suffer from the bug
indicated by Roger. OTOH, since it isn't supported by special code, it is not
as efficient as ,/ . But it's nearly as terse,
equally understandable, and is, until the next release, more reliable than its
optimized counterpart.
-Dan
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
