[R] Change global env variables from within a function

[Erik Iverson]

Hello,

Jon Zadra wrote:
>  Hi,
> 
> I've looked all over for a solution to this, but haven't had much look 
> in specifying what I want to do with appropriate search terms.  Thus I'm 
> turning to R-help.
> 
> In the process of trying to write a simple function to rename individual 
> column names in a data frame, I ran into the following problem: When I 
> rename the columns within my function, I can't seem to get it to apply 
> to the data frame in the global environment in a simple manner.

Not a direct answer to your question, but
does the ?rename function in the reshape package do what you want?

In general, the paradigm that I think R tries to promote is to
call functions that *return values* that you can assign to object
names, so calling "colrename" below to change an object in the
global environment is warned against for its side effects.

You could have the function return the modified data.frame, but
of course that's what the "rename" function I mention above does
already.

You might also see ?transform.

> 
> Given:
> 
> tempdf <- data.frame("a" = 1:6, "b" = 7:12)
> 
> #I can rename a to g this way:
> names(tempdf)[names(tempdf)=="a"] <- "g"
> 
> #Wanting to simplify this for the future, I have the function:
> colrename <- function(dframe, oldname, newname) {
>     names(dframe)[names(dframe)==oldname] <- newname
>     }
> 
> colrename(tempdf, "a", "g")
> 
> #However of course the change to tempdf stays within colrename().  I 
> could add "return(names(dframe))" to the function and then call the 
> function like:
> #names(tempdf) <- colrename(tempdf, "a", "g")
> #but at that point its not much simpler than the original:
> #names(tempdf)[names(tempdf)=="a"] <- "g"
> 
> So, i'm wondering, is there a way to write code within that function so 
> that the tempdf in the global environment gets changed?
> I thought of doing "names(dframe)[names(dframe)==oldname] <<- newname" 
> but then of course it doesn't replace "dframe" with "tempdf" in the 
> evaluation, so it seems to be a problem of looking at some variables 
> within the function environment but others in the parent environment.
> 
> Any help is greatly appreciated.
> 
> Thanks,
> 
> Jon