This is now fixed in R-devel (r 45953).
luke
On Mon, 14 Apr 2008, bill at insightful.com wrote:
> Full_Name: Bill Dunlap
> Version: 2.8.0 Under development (unstable) svn 45325
> OS: Linux
> Submission from: (NULL) (76.28.245.14)
>
>
> It is difficult to write wrapper functions
> for unix.time(expr) because it uses the idiom
> expr <- substitute(expr)
> eval(expr, envir=sys.parent())
> to evaluate the input expression. Here is an
> example of the problem
> > elapsed.time<-function(...)unix.time(...)[3]
> > sapply(1:3,
function(seconds.arg)elapsed.time(Sys.sleep(seconds.arg)))
> Error in Sys.sleep(seconds.arg) : object "seconds.arg" not
found
> Timing stopped at: 0.002 0 0.002
>
> I think thatif unix.time(expr) made use of lazy evaluation of
> arguments and used just
> expr
> at the point where it wanted expr to be evaluated,
> then the evaluation would take place in the right frame.
> If I make the following change
> diff -c /tmp/unix.time.R /tmp/my.unix.time.R
> *** /tmp/unix.time.R 2008-04-14 12:22:39.000000000 -0700
> --- /tmp/my.unix.time.R 2008-04-14 12:22:32.000000000 -0700
> ***************
> *** 10,23 ****
> }
> if (!exists("proc.time"))
> return(rep(NA_real_, 5))
> - loc.frame <- parent.frame()
> if (gcFirst)
> gc(FALSE)
> - expr <- substitute(expr)
> time <- proc.time()
> on.exit(cat("Timing stopped at:", ppt(proc.time() - time),
> "\n"))
> ! eval(expr, envir = loc.frame)
> new.time <- proc.time()
> on.exit()
> structure(new.time - time, class = "proc_time")
> --- 10,21 ----
> }
> if (!exists("proc.time"))
> return(rep(NA_real_, 5))
> if (gcFirst)
> gc(FALSE)
> time <- proc.time()
> on.exit(cat("Timing stopped at:", ppt(proc.time() - time),
> "\n"))
> ! expr # evaluate in its original frame by lazy evaluation
> new.time <- proc.time()
> on.exit()
> structure(new.time - time, class = "proc_time")
>
> then my wrapper function works as expected:
>> sapply(1:3, function(seconds.arg)elapsed.time(Sys.sleep(seconds.arg)))
> elapsed elapsed elapsed
> 1.001 2.001 3.001
>
> Another approach would be to find the environment that expr
> came from and feed that into eval(). Is there a way to get
> the environment that an unevaluated argument was created in?
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>
--
Luke Tierney
Chair, Statistics and Actuarial Science
Ralph E. Wareham Professor of Mathematical Sciences
University of Iowa Phone: 319-335-3386
Department of Statistics and Fax: 319-335-3017
Actuarial Science
241 Schaeffer Hall email: luke at stat.uiowa.edu
Iowa City, IA 52242 WWW: http://www.stat.uiowa.edu