R help - Feb 2021 - Is it Possible to Create S4 Function Objects?

Dear All,

I was wondering if it's possible to create S4 function objects?
(Or S4 closures, if you prefer).

i.e.
An R object, that is both an S4 object, and a function.

This would allow one to write:
> f <- constructor4.functionobj ()
> f ()
> f at slot.of.f
I've searched for this, but I can't find any examples.

If it's possible, then that leads to the question of if/how the body
of f() could directly access the value of the slot?
> f <- function ()
>    this at slot.of.f
I should note that the more common approach of storing values in a
function's environment, doesn't work well for top-level objects.
This approach is dependent on immutability of the function's data, and
I'm reluctant to create such constraints.

It appears that my previous post may lack clarity.

So, I've created three examples of mathematical function objects.
(One) using a standard closure with lexical scoping, (two) using S3
with an attribute and (three) using S4 with a slot.
Each example constructs the function object via a constructor (of some
form), plots it, then prints the attribute/etc.
However, the third example with S4 is incomplete.

Any suggestions welcome.

----begin code----

plotf <- function (f)
{   x <- seq (-1, 1,, 200)
    plot (x, f (x), type="l")
}

#standard closure, with lexical scoping
quad.lex <- function (p = c (0, 0, 1) )
{   function (x)
        p [1] + p [2] * x + p [3] * x^2
}

#s3-based function object, with attribute
quad.s3 <- function (p = c (0, 0, 1) )
{   f <- function (x)
    {   this <- sys.function ()
        p <- attr (this, "p")
        p [1] + p [2] * x + p [3] * x^2
    }
    attr (f, "p") <- p
    f
}

#s4-based function object, with slot
setClass ("Quad.S4", slots = list (p="numeric") )
Quad.S4 <- function (p = c (0, 0, 1) )
{   #?
}

f.lex <- quad.lex ()
plotf (f.lex)
environment (f.lex)$p

f.s3 <- quad.s3 ()
plotf (f.s3)
attr (f.s3, "p")

f.s4 <- Quad.S4 ()
#plotf (f.s4)
#f.s4 at p


On Sat, Feb 13, 2021 at 10:53 AM Abby Spurdle (/??bi/)
<spurdle.a at gmail.com> wrote:
>
> Dear All,
>
> I was wondering if it's possible to create S4 function objects?
> (Or S4 closures, if you prefer).
>
> i.e.
> An R object, that is both an S4 object, and a function.
>
> This would allow one to write:
>
> > f <- constructor4.functionobj ()
> > f ()
> > f at slot.of.f
>
> I've searched for this, but I can't find any examples.
>
> If it's possible, then that leads to the question of if/how the body
> of f() could directly access the value of the slot?
>
> > f <- function ()
> >    this at slot.of.f
>
> I should note that the more common approach of storing values in a
> function's environment, doesn't work well for top-level objects.
> This approach is dependent on immutability of the function's data, and
> I'm reluctant to create such constraints.

Hi Abby,

Something along the line of:

   setClass("S4Function",
     contains="function",
     representation(name="character", more_stuff="ANY")
   )

seems to do what you want:

   f <- new("S4Function", function(a) a^2, name="square")

   # 'f' is both an S4 object and a function:
   is.object(f)
   # [1] TRUE
   is.function(f)
   # [1] TRUE

   f at name
   # [1] "square"
   f(11)
   # [1] 121

Hope this helps,

H.

On 2/12/21 1:53 PM, Abby Spurdle (/??bi/) wrote:
> Dear All,
> 
> I was wondering if it's possible to create S4 function objects?
> (Or S4 closures, if you prefer).
> 
> i.e.
> An R object, that is both an S4 object, and a function.
> 
> This would allow one to write:
> 
>> f <- constructor4.functionobj ()
>> f ()
>> f at slot.of.f
> 
> I've searched for this, but I can't find any examples.
> 
> If it's possible, then that leads to the question of if/how the body
> of f() could directly access the value of the slot?
> 
>> f <- function ()
>>     this at slot.of.f
> 
> I should note that the more common approach of storing values in a
> function's environment, doesn't work well for top-level objects.
> This approach is dependent on immutability of the function's data, and
> I'm reluctant to create such constraints.
> 
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
> 
-- 
Herv? Pag?s

Bioconductor Core Team
hpages.on.github at gmail.com