[Rd] ** operator

[Prof Brian Ripley]

On Thu, 15 May 2008, Roger Bivand wrote:

> On Thu, 15 May 2008, Barry Rowlingson wrote:
>
>> Duncan Murdoch wrote:
>>
>>>  It's fairly unlikely to go away, but it's an old undocumented feature, so
>>>  use at your own risk.  There are lots of other undocumented features, but
>>>  maybe no others at the lexical level.  Check out src/main/gram.y if you
>>>  want to search for more at that level, and src/* if you want them at any
>>>  level.
>> 
>> 
>> I just had a quick browse in the source and I can't see where "**" is 
>> defined. names.c relates ^ to the power operation in arithmetic.c but no 
>> sign of "**":
>> 
>> ./names.c:{"^",         do_arith,       POWOP,  1,      2, {PP_BINARY2, 
>> PREC_POWER,  1}},
>> 
>> grepping for POWOP doesn't help.
>>
>>  It's not an operator like '*':
>>>  get("**")
>> Error in get("**") : variable "**" was not found
>>>  get("*")
>> function (e1, e2)  .Primitive("*")
>> 
>> and there's no ** in gram.y.
>> 
>> searching for 'power' and quoted "**" doesn't help me either. Searching for 
>> unquoted ** just produces too much C code to be useful.
>> 
>> It's got to be there somewhere!!!
>
> Is there a clue in:
>
>> "^"(5, 2)
> [1] 25
>> "**"(5, 2)
> Error: could not find function "**"
>> 5 ^ 2
> [1] 25
>> 5 ** 2
> [1] 25
>
> that is, it is truly obfuscated and not accessible using contemporary 
> interfaces?

AFAIK it has only ever been used by the parser, which replaces it by ^ 
in the tokenizer.  I believe it is undocumented in R, but S used to allow 
it (and may still), so presumably it is for compatibility.

As Duncan Murdoch says, use undocumented features at your own risk.  I've 
contemplated removing this one at times.

-- 
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