Seems I found a bug, though the following approach still works, and has the
best performance
ar =: 1 : '5!:1 <''u'''
the bug is:
((3 ar)`) (`(5 ar)) (`:6)
((3`)(`5))(`:6) NB. not valid gerunds, though magically the following still
works
+ ((3 ar)`) (`(5 ar)) (`:6)
8
+ A=: (((3 ar)`) (`(5 ar)) (`:6)) NB. separate lines still work despite
"invalid return"
8
though the following fails,
+ ((3`)(`5))(`:6)
|domain error
representation is supposed to be,
+ ((3 ar)`) (`(5 ar))
┌─────┬─┬─────┐
│┌─┬─┐│+│┌─┬─┐│
││0│3││ ││0│5││
│└─┴─┘│ │└─┴─┘│
└─────┴─┴─────┘
An incompatible change that would seem very welcome to me, would be that
noun ` noun
instead of creating a list from the 2 nouns (pair) if they are magically
compatible, would create a pair of atomic representations of each noun.
Because the above code works, it appears as though just "visual sugar" is
provided for display. The proposed incompatible change would make the visual
sugar valid code, which is consistent (other than this) within J.
On Thursday, February 27, 2020, 11:33:59 a.m. EST, Raul Miller
<rauldmil...@gmail.com> wrote:
Or, more robust:
C=:2 :0
mRep=. 5!:5<'m'
mVn=. m V n
mVnRep=. 5!:5<'mVn'
1 :('(',mRep,') u~ ',mVnRep)
)
Example use:
V=:+
,. 10 C (i.2 3)
10 11 12 10
13 14 15 10
(The issue here is that 5!:5 is better than ": when serializing nouns
for use in sentences.)
Perhaps even better, though, would be to define a verb to serialize
nouns and use that in place of ":
lrep=:3 :0
'(',(5!:5<'y'),')'
)
C=:2 :0
1 :((lrep m),'u~',lrep m V n)
)
Thanks,
--
Raul
On Thu, Feb 27, 2020 at 6:48 AM 'Pascal Jasmin' via Programming
<programm...@jsoftware.com> wrote:
>
> C =: 2 : 0
> 1 : ((": m) , ' u~ ' , ": m V n)
> )
>
>
>
>
>
>
> On Thursday, February 27, 2020, 03:04:57 a.m. EST, Hauke Rehr
> <hauke.r...@uni-jena.de> wrote:
>
>
>
>
>
> What I want is rather something (C here) callable like
> noun1 C noun2
> resulting in the adverb
> noun1 u~ <result of noun1 V noun2>
>
> What you say I knew except if the evaluation
> rules changed in j9 and this is new bahaviour.
>
> That said, I got rid of my nested 1 :/2 :
> construct (I didn’t mention that yet)
> and tried replacing m and n by x and y
> but to no avail.
>
> Am 27.02.20 um 01:37 schrieb 'Pascal Jasmin' via Programming:
> > u or m will be the argument to A. x and y arguments to resulting verb.
> >
> > it sounds as though you may be trying to do
> >
> > A =: 1 : 'x u~ x V y' NB. a valid dyadic adverb in j9
> >
> > and this would be equivalent to
> >
> > ([ u~ V)
> >
> >
> >
> >
> > On Wednesday, February 26, 2020, 06:20:45 p.m. EST, Hauke Rehr
> > <hauke.r...@uni-jena.de> wrote:
> >
> >
> >
> >
> >
> > Hello again,
> >
> > I’m still confused about modifiers.
> > Please tell me where to find more information
> > that might aid in understanding how this works:
> >
> > I have an adverb A and a verb V but sometimes
> > x and y for the derived verb for A are known
> > prior to the verb, so I want to write a modifier
> > something like this:
> >
> > m (u A)~ m V n
> >
> > I get a domain error which afaik is due to both
> > m and u being bound to the left noun.
> > So how would one go about this and where to find
> > further information on constructions like this?
> >
>
> --
> ----------------------
> mail written using NEO
> neo-layout.org
>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
