[R] Writing R package and do.call

[Rolf Turner]

On 3/03/2009, at 2:46 PM, Rob Foxall wrote:

> Dear R-Help,
>
> I have written a function, which in simplified format can be  
> represented as:
>
> temp.fn <- function(my.mean,my.sd){
>    Parameters <- list(mean = my.mean, sd = my.sd)
>    curve(do.call(dnorm,c(list(x), Parameters)), from = my.mean-my.sd,
> to = my.mean+my.sd)
> }
>
> This works as I want it do. Wishing to immortalise this function into
> my very own package however, results in the following warning:
> (Running R CMD check myPackage)
>
> * checking R code for possible problems ... NOTE
> temp.fn: no visible binding for global variable 'x'
>
> Although it doesn't seem to matter, I would feel more comfortable if
> there was an alternative way which avoided this warning.
> ( I can avoid the warning, by pointlessly defining x, e.g. "x <- 1",
> but this seems pretty ugly!)

Rob:  This is almost as ugly --- or even more ugly --- but it's  
sexier in that
it's more sophisticated.  (Nudge, nudge, wink, wink!)  Anyhow, I  
think you can
work around the warning as follows: Put the entire call to curve()  
inside a text
string, and then call ``eval() on that text string:

temp.fn <- function(my.mean,my.sd){
    Parameters <- list(mean = my.mean, sd = my.sd)
    xxx <-"curve(do.call(dnorm,c(list(x), Parameters)), from =  
my.mean-my.sd,to = my.mean+my.sd)"
    eval(parse(text=xxx))
}

This seems to run, and when I put temp.fn.R inside an arbitrary  
package and built that package,
it built without issuing a warning.

HTH.

	cheers,

		Rolf

######################################################################
Attention:\ This e-mail message is privileged and confid...{{dropped:9}}