R help - Nov 2007 - test for existance of a method for given class

Dear All,

I want to test whether a method exists for given object. For example, 
whether a function "deviance" is defined for an object of the
"lm" class.

My imperfect understanding leads me to think something like

  hasMethod("deviance", object)
  hasMethod("deviance", "lm")
  existsMethod("deviance", signature(class="lm"))

or similar might work (I don't fully understand how to manipulate 
signatures), but all the variations on this I have tried return FALSE. 
(Except, interestingly, when I first load library lme4, after which all 
return TRUE even for non-existant classes and functions).

I realize there are several ways in which R implements function 
polymorphism and that this is all documented somewhere but a hint would 
save me considerable time. I would also prefer not to resort to the hack 
solution of try()ing the function with the object and then catching the 
error to determine whether it was defined.

Thanks,

Will

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-Dr William Valdar   
++44 (0)1865 287 589
Wellcome Trust Centre           valdar at well.ox.ac.uk
for Human Genetics, Oxford      www.well.ox.ac.uk/~valdar

On Wed, 14 Nov 2007, William Valdar wrote:
> Dear All,
>
> I want to test whether a method exists for given object. For example,
> whether a function "deviance" is defined for an object of the
"lm" class.
For an S3 generic 'f' and with an S3 object or an S3 class 'x',
try

hasS3method <- function(f, x)
{
     if(is.object(x)) x <- oldClass(x)
     m <- methods(f)
     cl <- sub(paste("^", f, ".", sep=""),
"", m)
     any(c("default", x) %in% cl)
}

(You can break this, e.g. by f="resid" or using implicit classes: it
needs
inside knowledge to know if the latter would be invoked.  Also, the set 
of available methods is in principle scope-specific.)

For S4 generics and classes, look at selectMethod(optional=TRUE): this is 
documented to return NULL if and only if there is no applicable method.

> My imperfect understanding leads me to think something like
>
>  hasMethod("deviance", object)
>  hasMethod("deviance", "lm")
>  existsMethod("deviance", signature(class="lm"))
>
> or similar might work (I don't fully understand how to manipulate
> signatures), but all the variations on this I have tried return FALSE.
> (Except, interestingly, when I first load library lme4, after which all
> return TRUE even for non-existant classes and functions).
>
> I realize there are several ways in which R implements function
> polymorphism and that this is all documented somewhere but a hint would
> save me considerable time. I would also prefer not to resort to the hack
> solution of try()ing the function with the object and then catching the
> error to determine whether it was defined.
>
> Thanks,
>
> Will
>
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-> Dr William
Valdar               ++44 (0)1865 287 589
> Wellcome Trust Centre           valdar at well.ox.ac.uk
> for Human Genetics, Oxford      www.well.ox.ac.uk/~valdar
-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595