The R Project SVN R

Rev

Rev 2759 | Rev 7225 | Go to most recent revision | Blame | Compare with Previous | Last modification | View Log | Download | RSS feed

\name{eval}
\title{Evaluate an (Unevaluated) Expression}
\usage{
eval(expr, envir=sys.frame(sys.parent()), 
   enclos=if(is.list(envir) || is.pairlist(envir)) 
             sys.frame(sys.parent()))

evalq(expr, envir, enclos)

eval.parent(expr, n = 1)

local(expr, envir = new.env())
}
\alias{eval}
\alias{evalq}
\alias{eval.parent}
\alias{local}
\arguments{
  \item{expr}{object of mode \code{\link{expression}} or\code{call} or
      an ``unevaluated expression''.}
  \item{envir}{the \code{\link{environment}} in which \code{expr} is to be
      evaluated. May also be a list or an integer as in \code{sys.call}.}
  \item{enclos}{Only relevant if \code{envir} is a list. Specifies the
      enclosure, i.e. where R looks for objects not found in \code{envir}.} 
  \item{n}{parent generations to go back}
}
\description{
These functions evaluate the expression \code{expr} argument in the
environment specified by \code{envir} and returns the computed value.  If
\code{envir} is not specified, then
\code{\link{sys.frame}(\link{sys.parent}())}, the environment where the
call to \code{eval} was made is used. The \code{eval} form evaluates its first
argument before passing it to the evaluator.
This allows you to assign complicated expressions to symbols and then
evaluate them. The \code{evalq} form avoids this and is equivalent to
\code{eval(quote(expression), ...)} 

\code{eval.parent(expr, n)} is a shorthand for \code{eval(expr,
parent.frame(n))}. 

\code{local} evaluates an expression in a local environment. It is
equivalent to \code{evalq} except the its default argument creates a
new, empty environment. This is useful to create anonymous recursive
functions and as a kind of limited namespace feature since variables
defined in the environment are not visible from the outside. 
}


\seealso{
\code{\link{expression}}, \code{quote}, \code{\link{sys.frame}},
\code{\link{environment}}. 
}
\note{
Due to the difference in scoping rules, there are some differences
between R and S in this area. In particular, the default enclosure
in S is the global environment.

When evaluating expressions in dataframes that has been passed as
argument to a function, the relevant enclosure is often the caller's
environment, i.e. one needs 
\code{eval(x, data, sys.frame(sys.parent()))}.
}

\examples{
eval(2 ^ 2 ^ 3)
mEx <- expression(2^2^3); mEx; 1 + eval(mEx)
eval({ xx <- pi; xx^2}) ; xx

a <- 3 ; aa <- 4 ; evalq(evalq(a+b+aa, list(a=1)), list(b=5)) # == 10
a <- 3 ; aa <- 4 ; evalq(evalq(a+b+aa, 1), list(b=5))         # == 12

ev <- function() {
    e1 <- sys.frame(sys.parent())
    ## Evaluate a in e1
    aa <- eval(expression(a),e1)
    ## evaluate the expression bound to a in e1
    a <- expression(x+y)
    list(aa = aa, eval = eval(a, e1))
}
tst.ev <- function(a = 7) { x <- pi; y <- 1; ev() }
tst.ev()#-> aa : 7,  eval : 4.14

##
## Uses of local()
##

# Mutual recursives. 
# gg gets value of last assignment, an anonymous version of f.

gg<-local({
    k<-function(y)f(y)
    f<-function(x)if(x)x*k(x-1)else 1
})
gg(10)
sapply(1:5,gg)

# Nesting locals. a is private storage accessible to k
gg<-local({
    k<-local({
        a<-1
        function(y){print(a<<-a+1);f(y)}
    })
    f<-function(x)if(x)x*k(x-1)else 1
})
sapply(1:5,gg)

ls(envir=environment(gg))
ls(envir=environment(get("k",envir=environment(gg))))

}

\keyword{data}
\keyword{programming}