[R] stumped by eval

[Peter Dalgaard]

Ross Boylan wrote:
> In the following example, the inner evaluation pulls in the global value
> of subset (a function) rather than the one I thought I was passing in (a
> vector).  Can anyone help me understand what's going on, and what I need
> to do to fix the problem?
>
> f0 <- function(formula, data,
>                               subset, na.action
>                 )
> {
>   f1(formula, data,
>                      subset, na.action
>                      )
> }
>
>
> f1 <- function (formula, data,
>                               subset, na.action
>                 )
> {
>     mf <- match.call()
>     mf[[1]] <- as.name("model.frame")
>     mf <- eval(mf, parent.frame())
>   }
>
>   
>> netto <- data.frame(Reading=seq(3), Spec=seq(3), Reader=seq(3))
>> t3 <- f0(Reading~0+Spec+Reader, netto, c(1))
>>     
> Error in xj[i] : invalid subscript type 'closure'
>   
>> traceback()
>>     
> 7: `[.data.frame`(list(Reading = 1:3, Spec = 1:3, Reader = 1:3), 
>        function (x, ...) 
>        UseMethod("subset"), , FALSE)
> 6: model.frame.default(formula = formula, data = data, subset = subset, 
>        na.action = na.action)
> 5: model.frame(formula = formula, data = data, subset = subset, 
>        na.action = na.action)
> 4: eval(expr, envir, enclos)
> 3: eval(mf, parent.frame())
> 2: f1(formula, data, subset, na.action)
> 1: f0(Reading ~ 0 + Spec + Reader, netto, c(1))
>
> I started with a case in which f0 was called with only 2 arguments
> (i.e., subset was missing), and that is the case I'm ultimately
> interested in.  However, even the situation above is failing.
>
> According to resume, the class of subset is numeric in frames 2 and 4,
> but a function in frames 3 and 5.  That may be correct for frame 3,
> since it is the evaluation function without a named argument subset (and
> thus it picks up the global value).  But it's wrong for 5 (and higher).
>
> eval itself is primitive; I don't understand exactly what is going on
> with the 2 entries for it (frames 3 + 4).
>
>   
Yes, this is elusive, but it is not actually eval() that is doing you 
in. It is the notion of a model environment, see the last bit of ?formula.

The point is that subset (and offset) arguments are subject to the same 
evaluation rules as the terms inside the formula: First look in "data", 
then in the environment of the formula, which in this case is the global 
environment.

This behaviour is generally a good thing because it prevents you from 
accidentally picking up internal variables of f1, but working around it 
can be a little painful. As far as I recall, you can use an explicit 
substitute of the subset argument.

> R package 2.6.1-1 on Debian GNU/Linux.
>
> ______________________________________________
> 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.
>   


-- 
   O__  ---- Peter Dalgaard             Øster Farimagsgade 5, Entr.B
  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
 (*) \(*) -- University of Copenhagen   Denmark          Ph:  (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)                  FAX: (+45) 35327907