Re: [R] (not so real) function (more like #include)

From: Gabor Grothendieck <ggrothendieck_at_gmail.com>
Date: Thu 05 Oct 2006 - 16:54:54 GMT

I thought of a few more possibilities.

If options 1, 2 and 3 correspond to longfun, longfun2 and longfun3 then:

4. We could arrange it so that the variables are stored outside of longfun and then longfun, f1 and f2 reference them symmetrically.

a <- b <- d <- 1
longfun4 <- function() {

   out <- f1()
   # since a is not in longfun4 use <<- to set it    a <<- a + 1
   f2() + out
}
f1 <- f2 <- function() a + b + d
longfun4()

Given the problem as stated I don't think this approach is really the best but if the problem gets extended to one where there is not just one longfun but several longfuns then it will not be feasible to store all the variables in longfun since the other longfuns won't be able to access them and this approach becomes desirable.

5. We could also use the proto package for this purpose placing a, b, d, longfun, f1 and f2 into an object p. If we wanted to be able to override f1, f2, a, b and d in delegated subobjects and be able to refer to f1 and f2 outside of longfun, i.e. outside of the context of object p, then we would have to refer to them as .$f1, .$f2, .$a, .$b and .$d and also pass the target object as arg 1 of f1 and f2 but we have not bothered with that here since it seems not to be a requirement:

library(proto)
p <- proto(a = 1, b = 1, d = 1)
p$longfun <- function(.) {

   out <- f1()
   # since a is not in longfun4 use <<- to set it    a <<- a + 1
   f2() + out
}
p$f1 <- p$f2 <- function() a + b + d
p$longfun() # 7

This is probably overkill here but it might be that if one thinks about the design some more that there are some opportunities to define subobjects of p in which case the ability to do inheritance would come into play.

6. I think the R.oo package would work too although I am not sure its as suitable here since it would require the definition of classes which you don't really need. proto lets you define objects directly even without classes.

On 10/5/06, Meinhard Ploner <meinhardploner@gmx.net> wrote:
> Thanks a lot! longfun2 & longfun3 work perfect for my "very
> untypical" problem. As I have many local variables, usual functions
> with parameters are very uncomfortable, but the code you gave me is
> great!
> Meinhard
>
>
> On Oct 4, 2006, at 5:25 PM, Gabor Grothendieck wrote:
>
> > longfun could just pass a, b and d to each of the individual
> > functions and each of the individual functions could pass
> > out back as a return value.
> >
> > f1 <- f2 <- function(a, b, d) a+b+d
> >
> > longfun1 <- function() {
> > a <- b <- d <- 1
> > out <- f1(a, b, d)
> > out <- f2(a, b, d) + out
> > out
> > }
> >
> > longfun1() # 6
> >
> > If the problem is that a, b and d actually represent 100 variables,
> > say,
> > and its a nuisance to pass them all explicitly you could do this:
> >
> > f1 <- f2 <- function() with(parent.frame(), a + b + d)
> >
> > longfun2 <- function() {
> > a <- b <- d <- 1
> > out <- f1()
> > out <- f2() + out
> > out
> > }
> >
> > longfun2() # 6
> >
> >
> > or you could do it this way if you would prefer to have longfun
> > control the scoping rather than having it done within each
> > individual function:
> >
> >
> > f1 <- f2 <- function() a+b+d
> >
> > longfun3 <- function() {
> > a <- b <- d <- 1
> > environment(f1) <- environment(f2) <- environment()
> > out <- f1()
> > f2() + out
> > }
> >
> > longfun3() # 6
> >
> >
> > On 10/4/06, Meinhard Ploner <meinhardploner@gmx.net> wrote:
> >> Hello R users,
> >>
> >> I have a very long function with parts of that looking like a list of
> >> jobs. The first part of the function defines a lot of local
> >> variables, which are used by the jobs. Now I extracted the "job" part
> >> und putted them into an external file, used by "source(, local=T)",
> >> which is more comfortable to me. Apart from that it gives me the
> >> opportunity that more users can work on the project. Is it possible
> >> to define a function for that code and passing to it the environment
> >> of f() without save(list=ls()) and load() .... ?
> >>
> >> Thanks in advance
> >> Meinhard Ploner
> >>
> >> longfun <- f() {
> >>
> >> ## lot of local variables:
> >> a <- ...
> >> b <- ...
> >> d <- ...
> >> ...
> >>
> >> out <- ...
> >>
> >> source("job1.txt", local=TRUE) #changes out, uses a, b,
> >> d, ...
> >>
> >> source("job2.txt", local=TRUE) # changes out, uses a, b,
> >> d, ...
> >> ...
> >> }
> >>
> >>
> >>
> >> version
> >> _
> >> platform i386-apple-darwin8.6.1
> >> arch i386
> >> os darwin8.6.1
> >> system i386, darwin8.6.1
> >> status
> >> major 2
> >> minor 3.1
> >> year 2006
> >> month 06
> >> day 01
> >> svn rev 38247
> >> language R
> >> version.string Version 2.3.1 (2006-06-01)
> >>
> >> ______________________________________________

> >> R-help@stat.math.ethz.ch mailing list
> >> 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.
> >>
> >
>
>



R-help@stat.math.ethz.ch mailing list
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. Received on Fri Oct 06 02:57:51 2006

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.1.8, at Thu 05 Oct 2006 - 17:30:08 GMT.

Mailing list information is available at https://stat.ethz.ch/mailman/listinfo/r-help. Please read the posting guide before posting to the list.