I think the motivation is that N0 (A1 A2) just oughta behave like ((N0
A1) A2) because, well, what else would it do? It does behave that way
when the result R3 is noun/verb/conjunction (with conjunction giving an
error); why not adverb? Methinks the current behavior is simply a bug.
If it turned out to be difficult - if creating the composite (R3 A2)
posed problems - I would rethink.
Henry Rich
On 3/15/2016 6:12 PM, Raul Miller wrote:
So, ok...
Adverb *trains* get created by "6 Bident".
Adverb *application* gets handled by "3 Adverb".
And, yes, this includes the application of adverb trains.
And, yes, the dictionary's coverage of the behavior adverb trains is
pretty much just a few examples.
And, more generally, error cases can be reimplemented to do something
other than produce an error. There's some room for small bits of
linguistic drift.
That said, there's the question of usefulness. If you defined adverb
train behavior such that v (A1 A2) where v A1 produces an adverb to be
an adverb result of the form ((v A1) A2), we could do that, and that
would prevent the error from being a syntax error. But that would also
mean that if someone accidentally wrote (A1 A2) but meant to write
something else they would not get an error. But both of those seem to
be rather unlikely.
So, are there any motivating useful examples which would make this
particular change worth including in the official interpreter?
(And, yes, that is a really tough question. But I think it's a fair question.)
Thanks,
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
