[R] How '.' is used?

[Duncan Murdoch]

(Ted Harding) wrote:
> On 09-Aug-09 16:53:32, Douglas Bates wrote:
>   
>> On Sun, Aug 9, 2009 at 11:32 AM, Ted
>> Harding<Ted.Harding at manchester.ac.uk> wrote:
>>     
>>> On 09-Aug-09 16:06:52, Peng Yu wrote:
>>>       
>>>> Hi,
>>>> I know '.' is not a separator in R as in C++. I am wondering where it
>>>> discusses the detailed usage of '.' in R. Can somebody point me a
>>>> webpage, a manual or a book that discuss this?
>>>>
>>>> Regards,
>>>> Peng
>>>>         
>>> To the best of my knowledge, apart from its specific use as a
>>> separator
>>> between the integer and fractional parts of a number, "." has no
>>> specific
>>> use in R, and you can, for instance, use it just as you would use an
>>> alphanumeric character in a name.
>>>
>>> For instance, you could do
>>>
>>> _. <- 1.2345
>>> _.
>>> _# [1] 1.2345
>>>
>>> _. <- function(x) x^2
>>> _.(12)
>>> _# [1] 144
>>>
>>> So, unless there is something I don't know about, there is hardly
>>> anything to discuss about "the detailed usage of '.' in R"!
>>>       
>> The ',' character is one of the characters allowed in names, hence it
>> can be used as you have suggested.  There are (at least) two special
>> usages of the '.' in names.  Following the time-honoured Unix
>> convention, names that begin with '.' are considered "hidden" names
>> and not listed by ls() or objects() unless you set all.names = TRUE in
>> the call.  Because of this convention it is inadvisable to use names
>> starting with '.' except when you wish to avoid potential name
>> conflicts.  The second special use of '.' in a name is in the
>> construction of the names of S3 method functions.  The method for
>> generic function "foo" applied to class "bar" is named "foo.bar".
>>     
>
> As in "summary.glm", I suppose! However, this prompts a question.
> In the first place, the construction of "summary.glm" from "summary"
> and "glm" is, in the first instance, simply using "." in its basic
> role as a permissible character in a name. Correct?
>   

Correct.
> Next -- and this is the real question -- how does R parse the name
> "summary.glm"? In my naivety, I simply suppose that it looks for
> an available function whose name is "summary.glm" in just the
> same way as it looks for "stopifnot", or for that matter "data.matrix"
> which is not (as far as I know) a compound of a generic function
> "data" applied to a class "matrix". Then "." would not have a special
> (parseable) role in the name -- it is simply another "letter".
>   

It doesn't do anything special when parsing.  The special sauce comes 
when the generic summary() function executes UseMethod("summary").  At 
that point, we know the class of the object, and we know the name of the 
generic, so it goes looking for a summary.glm method.

There are some subtleties to how it does that lookup (see the discussion 
in Writing R Extensions about NAMESPACES), but if you had a generic 
function calling UseMethod("data") and it was passed an object of class 
"matrix", data.matrix() would be called, even though that doesn't make 
sense.  This is a flaw in the S3 system, and one of the motivations for 
the development of the S4 system.

Duncan Murdoch
> But when you do have such a function, like "summary.glm", does R
> in fact parse it as "summary" then "glm" (i.e. look out for the
> generic function "summary" and then specialise it to handle "glm").
>
> As I say, I suppose not. And, if not, then the "special use" of
> the character "." is simply a programmer's convention for the
> construction of the name, and once the name exists the "." does
> not have a special (parseable) significance for R.
>
> Just seeking clarification ... !
> Thanks,
> Ted.
>
> --------------------------------------------------------------------
> E-Mail: (Ted Harding) <Ted.Harding at manchester.ac.uk>
> Fax-to-email: +44 (0)870 094 0861
> Date: 09-Aug-09                                       Time: 19:58:32
> ------------------------------ XFMail ------------------------------
>
> ______________________________________________
> 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.
>