[Rd] possible bug in formals

[Prof Brian Ripley]

On Mon, 24 May 2010, Josef Leydold wrote:

> Dear Brian and Uwe,
>
> Thanks a lot for the clarification.
> I made the naive assumption that numeric constants in R are similar to
> those in C.
>
> Two questions still remain:
>
> (1) when I have a function
>
>    f<- function(a=1,b=-1) { a+b }
>
>    is it safe to use
>
>    val <- as.character(deparse(formals(f)$b))
>
>    to obtain a string that contains the default value for
>    argument "b". (Does is also work for other arguments with some
>    default of arbitrary class?)

But the defualt value is not character, so this cannot in general be 
done.

>
> (2) I have seen that packages like gWidget (in function ggenericwidget)
>    use a statement like
>
>    switch(class(formals(f)$b),
> 	numeric = { .... },
>        character = { .... },
>        class = { .... }, ....
>
>    for automatically processing function arguments.
>    in the case of "b=-1" this procedure obviously fails.
>    (I found this behavior of 'formals' while playing around with
>    packages "gWidgets" and "fgui" from CRAN).
>
>    Is there a safe workaround for this problem?
>
>    That is, is there a safe function that returns class
>    "numeric" for an exresion like "-1" or "-Inf"?

Why are you using class() when you seem to mean typeof()?

But the short answer is you appear to be trying to circumvent the 
language, and who really cares what the default is?  If it is used, it 
is evaluated, and then you can simply do typeof(b).  And if it is not 
used, who cares what it is?

>
> Josef
>
>
> On Sun, May 23, 2010 at 03:52:00PM +0100, Prof Brian Ripley wrote:
>> Documented too: from ?NumericConstants
>>
>>       Note that a leading plus or minus is not regarded by the parser as
>>       part of a numeric constant but as a unary operator applied to the
>>       constant.
>>
>>
>>
>> On Sun, 23 May 2010, Uwe Ligges wrote:
>>
>>>
>>>
>>> On 23.05.2010 16:14, Josef Leydold wrote:
>>>> Hi,
>>>>
>>>> I am a little bit surprised by the following output of
>>>> 'formals'. Is this the intended behavior?
>>>>
>>>>> f<- function(a=1,b=-1) { a+b }
>>>>> class(formals(f)$a)
>>>> [1] "numeric"
>>>>> class(formals(f)$b)
>>>> [1] "call"
>>>>
>>>>
>>>> Josef
>>>>
>>>>
>>>
>>>
>>> Yes, the arguments have not yet been evaluated, hence -1 is still a language
>>> object.
>>>
>>> Try to write
>>> f<- function(a= +1, b= -1) { a+b }
>>> and you will find that this is a fascinating feature.
>>>
>>> Uwe Ligges
>>>
>>> ______________________________________________
>>> R-devel at r-project.org mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>>
>>
>> --
>> 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
>>
>
> -- 
>
>
> -----------------------------------------------------------------------------
> Josef Leydold   |  WU (Vienna University of Economics and Business)
>                |  Institute for Statistics and Mathematics
> -----------------------------------------------------------------------------
> Augasse 2-6     |  Tel.   +43 1 31336 4695
> A-1090 Vienna   |  FAX    +43 1 31336 774
> European Union  |  email  josef.leydold at wu.ac.at
> -----------------------------------------------------------------------------
> Alles Unglueck kam daher, dass die Denkenden nicht mehr handeln konnten,
> und die Handelnden keine Zeit mehr fanden zu denken.       (Marlen Haushofer)
>
>

-- 
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