Yes Roger,  I mentioned in the first post of this thread that both use
ad-hoc encodings to implement anonymous recursions (and I was not very
proud of the tacit version).


On Sunday, November 18, 2018, roger stokes <[email protected]> wrote:

> Pepe, have you seen the Rosettacode entry for Y-combinator in J ?
> The tacit version is just before this:
>
>
> http://www.rosettacode.org/wiki/Y_combinator#Explicit_
> alternate_implementation
>
>  Regards
>
> On Fri, Nov 16, 2018 at 10:06 PM Jose Mario Quintana <
> [email protected]> wrote:
>
> > Yes, it is cheating.  :)  The problem is that the only names allowed
> > should refer to arguments (a la λ-calculus) and once you replace AAr the
> > assignment is visible.
> >
> > I think I figured out a way but I have tested it only a little bit,
> >
> >    M=. 'if. * y do. y * u <: y else. 1 end.' (1 :)
> >    M=. (5!:1)<'M'
> >
> >    Y=. (1 : '<(<,'':''),<(<(,''0'');1),<(,''0'');, (''u
> > u`:6('',(5!:5<''u''),'')`:6 y'')')(1 : 'u (u`:6)')
> >
> >    M Y("0) i.11
> > 1 1 2 6 24 120 720 5040 40320 362880 3628800
> >
> > It looks ugly but I guess that is the price one has to pay to be orthodox
> > and fully functional.
> >
> >
> >
> > On Fri, Nov 16, 2018 at 12:22 AM 'Pascal Jasmin' via Programming <
> > [email protected]> wrote:
> >
> > > I can cheat, by hiding the assignment in another function.
> > >
> > > AAr =: 1 : '(5!:1 < ''a'') a =. 1 : m'
> > >
> > > 'u u`:6('',(5!:5<''u''),'')`:6 y' AAr
> > > <(<,':'),<(<(,'0');1),<(,'0');,:'u u`:6('',(5!:5<''u''),'')`:6 y' (1
> : 'u
> > > u`:6('',(5!:5<''u''),'')`:6 y')
> > >
> > > this is not the same as Y though.
> > >
> > > but this is,
> > >
> > > YAr =: 1 : '1 : (''(5!:1 < ''''a'''') a =. 1 : '' , ''('' ,  m , '')''
> )'
> > > Y1 =. '''u u`:6('',(5!:5<''u''),'')`:6 y''' YAr
> > >
> > >
> > >
> > >
> > > ________________________________
> > > From: Jose Mario Quintana <[email protected]>
> > > To: [email protected]
> > > Sent: Thursday, November 15, 2018 6:01 PM
> > > Subject: Re: [Jprogramming] Revisisting the Y combinator
> > >
> > >
> > >
> > > Some authors regard functional programming as programming without any
> > > assignments.
> > >
> > > The wicked tacit fixed version of Y has no assignments whatsoever but
> it
> > is
> > > produced and it works by means of non-standard J code.  I could alter
> > > slightly the non-tacit version,
> > >
> > >    Y=. '(5!:1<''v'')v=. 1 : (''u u`:6('',(5!:5<''u''),'')`:6 y'')'(1 :)
> > >
> > > and get rid of its single assignment using a bit of wicked code but
> that
> > > defeats my purpose.
> > >
> > > I cannot figure out how to produce an orthodox version of Y with no
> > > assignments; then again, Y, actually, XY was the most complicated
> > non-tacit
> > > entity I have produced in a couple of decades and my failure probably
> > means
> > > little.
> > >
> > > I would be very interested to see an orthodox version of Y with no
> > > assignments if anyone can produce one.
> > >
> > > ----------------------------------------------------------------------
> > > 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
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to