[R] variable scope for deltavar function from emdbook

[Uwe Ligges]

This is R, not S-Plus.

In the first two lines you have

  expr <- as.expression(substitute(fun))
  nvals <- length(eval(expr, envir = as.list(meanval)))


Simplified example:

y <- 0

fn1 <- function(){
   y <- 1
   fn1sub <- function() print(y)
   fn1sub()
}


fn2sub <- function() print(y)
fn2 <- function(){
   y <- 2
   fn2sub()
}

fn1() #  1
fn2() #  0


So, if you do not want to pass the objects, but just use the symbols, 
you have to make sure the objects go into the environment you are 
evaluating the stuff in, or you have to ask the user to pass these 
objects to deltavar in another way.

This is a similar region of the hell as evaluating glm() within a 
function and using its formula interface.

Best,
Uwe Ligges







On 11.10.2011 14:09, Ben Bolker wrote:
> adad<adad<at>  gmx.at>  writes:
>
>> Working example:
>> ----------------------
>> library("emdbook")
>>
>> fn<- function()
>> {
>> browser()
>> y<- 2
>> print(deltavar(y*b2, meanval=c(b2=3), Sigma=1) )
>> }
>>
>> x<- 2
>> print(deltavar(x*b1, meanval=c(b1=3), Sigma=1) )
>> y<-3
>>
>> fn()
>> ------------------------
>>
>> running this returns 4 for the first function call, which is fine.
>>
>> For the call of deltavar in fn(), I get 9, i.e. the function uses y<-3
>> instead of the local y<-2. If y<- is commented, deltavar returns an error.
>>
>> So why is the function not using the local variable and how do I make it
>> use it?
>
>    The real problem is that I (the author) don't understand scoping in R, and
> how to manipulate it, as well as I'd like to. I will work on this (any
> tips from the R-helpers appreciated).  In the meantime, you could
> try out one of the other available delta-method calculators, such
> as the one in the msm package (library("sos"); findFn("{delta method}")).
>
>    More text to try to make gmane happy
>
>    Ben Bolker
>
> ______________________________________________
> R-help at r-project.org 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.