Pascal Jasmin wrote:
> 'mymethod' OO o should return the function mymethod__o
> (with o fixed reference)
Here's a way of accomplishing that. v1 produces a monad,
v2 produces a dyad:
refname=:4 :0
NB. x - name of member
NB. y - locale
nm=. x,'_',(":>y),'_'
)
v1=:2 :0
NB. deferred monad
if. 0=4!:0<nm=.m refname n do.
". bind nm
else.
a:1 :(nm,'@]')
end.
)
v2=:2 :0
NB. deferred dyad
if. 0=4!:0<nm=.m refname n do.
". bind nm
else.
a:1 :('[',nm,']')
end.
)
To defer resolution of the name, I'm creating a short
phrase that uses it. That said, your suggestion, below,
points at another possible approach:
> btw, a potential bug is that 5!:0 doesn't seem to work on a
> local variable that holds a name.
More specifically, names are resolved.
a=:+
c=:a_base_
c
a_base_
5!:1<'c'
+-------+
|a_base_|
+-------+
(5!:1<'c')5!:0
+
However, introducing an extra level of indirection lets us
work around this limitation:
b=:a_base_
c=:b
(5!:1<'c')5!:0
a_base_
By definition, (5!:1<'c')5!:0 should produce the same thing
as c, and in the context of locales it's clear that this is
not the case -- so I think this is a bug, and I think that
in some future version of J we might see this fixed.
--
Raul
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
