R devel - Mar 2013 - Ref classes initFields has incorrect environment?

As suggested by the following code:

A <- setRefClass("A", fields = list(x = "logical"),
  methods = list(initialize = function(...) {
    x <<- FALSE
    initFields(...)
}))
A$new()
# Works as expected

quote <- as.character
A$new()
# Error in function (value)  :
#  invalid replacement for field ?x?, should be from class ?logical?
or a subclass (was class ?character?)

I get the same error in:
* R version 2.15.2 (2012-10-26)
* R version 2.15.3 (2013-03-01)
* R Under development (unstable) (2013-03-28 r62432)

Hadley

-- 
Chief Scientist, RStudio
http://had.co.nz/

Nothing to do with initFields.  If you trace your redefined quote(), it's
called from the <<- assignment of x.

The "x" element in the environment for the reference class object is
implemented as an active binding in order to enforce the class when assigning
the field.

Effectively that makes the assignment behave like a reference class method, and
so ends up doing as() and getting back to the malware version of quote().

The fix will be to  have a more bullet proof (and perhaps more efficient)
version of the active binding generated for fields.  A better design anyway, but
this will take a little re-organization because the current default binding
function is generated by some code manipulation.

Meanwhile the workaround is: Don't do what the revised subject heading says.

John


On Mar 28, 2013, at 11:40 AM, Hadley Wickham <h.wickham at gmail.com>
wrote:
> As suggested by the following code:
> 
> A <- setRefClass("A", fields = list(x = "logical"),
>  methods = list(initialize = function(...) {
>    x <<- FALSE
>    initFields(...)
> }))
> A$new()
> # Works as expected
> 
> quote <- as.character
> A$new()
> # Error in function (value)  :
> #  invalid replacement for field ?x?, should be from class ?logical?
> or a subclass (was class ?character?)
> 
> I get the same error in:
> * R version 2.15.2 (2012-10-26)
> * R version 2.15.3 (2013-03-01)
> * R Under development (unstable) (2013-03-28 r62432)
> 
> Hadley
> 
> -- 
> Chief Scientist, RStudio
> http://had.co.nz/
> 
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel