R devel - Jun 2012 - "Incompatible methods" for overloaded operator

I'm trying to overload an operator, and I'm running into a strange
problem.
It happens when I install and load the package, but not when I simply
source() the code.

I'm defining + for two classes. The R code looks like this:
#' @export
#' @method "+" a
`+.a` <- function (x1, x2) {
    message("Running custom + function")
}

#' @export
#' @method "+" b
`+.b` <- `+.a`

In some cases I do a+b, and in other cases, I do b+b. I'm told that the +.a
and +.b functions must be identical to avoid the error about "Incompatible
methods". (In the actual code, the overloaded + function checks the classes
of x1 and x2, and then sends them off to other functions.)

This is the NAMESPACE file:
S3method("+",a)
S3method("+",b)

I've put the code up at https://github.com/wch/badadd.



If I just cut and paste the function definitions to my R session, it works
fine:
x + y
# Running + function
# NULL


However, if I install and load the package, it gives a warning about
incompatible methods, and then seems to fall back to the arithmetic +
operator:
library(badadd)
x + y
# [1] 3
# attr(,"class")
# [1] "a"
# Warning message:
# Incompatible methods ("+.a", "+.b") for "+"


Is this expected behavior? And if so, is there a workaround?
-Winston

	[[alternative HTML version deleted]]

Hi all,

Any ideas about this?  As far as I can tell it should work - and I
don't understand why it's ok when run outside of a package.

Hadley

On Wed, Jun 13, 2012 at 7:41 PM, Winston Chang <winstonchang1 at
gmail.com> wrote:
> I'm trying to overload an operator, and I'm running into a strange
problem.
> It happens when I install and load the package, but not when I simply
> source() the code.
>
> I'm defining + for two classes. The R code looks like this:
> #' @export
> #' @method "+" a
> `+.a` <- function (x1, x2) {
> ? ?message("Running custom + function")
> }
>
> #' @export
> #' @method "+" b
> `+.b` <- `+.a`
>
> In some cases I do a+b, and in other cases, I do b+b. I'm told that the
+.a
> and +.b functions must be identical to avoid the error about
"Incompatible
> methods". (In the actual code, the overloaded + function checks the
classes
> of x1 and x2, and then sends them off to other functions.)
>
> This is the NAMESPACE file:
> S3method("+",a)
> S3method("+",b)
>
> I've put the code up at https://github.com/wch/badadd.
>
>
>
> If I just cut and paste the function definitions to my R session, it works
> fine:
> x + y
> # Running + function
> # NULL
>
>
> However, if I install and load the package, it gives a warning about
> incompatible methods, and then seems to fall back to the arithmetic +
> operator:
> library(badadd)
> x + y
> # [1] 3
> # attr(,"class")
> # [1] "a"
> # Warning message:
> # Incompatible methods ("+.a", "+.b") for "+"
>
>
> Is this expected behavior? And if so, is there a workaround?
> -Winston
>
> ? ? ? ?[[alternative HTML version deleted]]
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel


-- 
Assistant Professor / Dobelman Family Junior Chair
Department of Statistics / Rice University
http://had.co.nz/

Hi,

I may be mistaken here but a possible interpretation of:

"If a method is found for just one argument or the same method is found 
for both, it is used."

could be that "the same method" here does not mean identical in term
of
object/content (as identical tests), but means that the two arguments 
resolve to the same method definition, i.e. a single common method 
exists for both arguments. This would happen is if a single method is  
defined for a common parent class. Try:


`+.C` <- function(x1, x2){
     message("Running custom + function")
}

x <- structure(1, class=c('A', 'C'))
y <- structure(1, class=c('B', 'C'))

x + y


Another work around would be to define only one method for class 'a' and
do the dispatch to `plus.b` in there (might even be able to call it 
`+.b` if not exported).

Might be a good suggestion to add a test in UseMethod to check is the 
methods are in fact identical. But there might be undesired side 
effects: two identical methods in different environment that behave 
differently depending on the state of the environment?
By the way any way of seeing any code of primitive function or are these 
pure C functions, which need to be searched for in R source code?

Renaud

-- 
Renaud Gaujoux
Computational Biology - University of Cape Town
South Africa