(Sorry, finger slipped early before.)
Ambrus wrote:
> I'd prefer (f :: n) to be
> defined as (f :: (n"f))
Roger responded:
> Not bad. The remaining thing
> to do about this, is to think
> about other extensions which
> may be possible to do with f :: n,
> which would be pre-empted with
> the proposed change.
One idea that springs to mind easily is gerunds. That is, n is a gerund,
and f :: n has some useful behavior, as with ^: and } [1]. Or f :: n
could mean "when f fails on input y, call it again, except with input n"
[2].
Of course, we'd have to come up with some very compelling uses which today's
f :: g syntax doesn't already admit (easily), to overcome the motivating
proposal f :: (n"f) .
-Dan
[1] One possible definition could be f :: gerund <=> f :: (ger...@.(13!:11
@: (''"_) , that is selexecuting from the gerund based on the error code f
produces .
[2] As in f :: ( f@:(n"_) ) .
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
