[Rd] precedence (was 'historical NA question')

[Hervé Pagès]

On 05/07/2014 02:55 PM, Simon Urbanek wrote:
>
> On May 7, 2014, at 5:41 PM, Hervé Pagès <hpages at fhcrc.org> wrote:
>
>> On 05/07/2014 02:01 PM, peter dalgaard wrote:
>>>
>>> On 07 May 2014, at 21:52 , Simon Urbanek <simon.urbanek at r-project.org> wrote:
>>>
>>>> On May 7, 2014, at 3:37 PM, Hervé Pagès <hpages at fhcrc.org> wrote:
>>>>
>>>>> No big deal. These things can be tricky:
>>>>>
>>>>> https://stat.ethz.ch/pipermail/r-devel/2006-January/036022.html
>>>>>
>>>>> Sorry I couldn't resist ;-)
>>>>>
>>>>
>>>> Yeah, but that's just yet another trip down the rabbit hole - why is -2 parsed as `-`(2) and not a single constant? Is there a way to express a negative constant in R? Hmm…
>>>>
>>>
>>> It's painful, but
>>>
>>>> bquote(.(-2)^2)
>>> -2^2
>>>> eval(bquote(.(-2)^2))
>>> [1] 4
>>>> bquote(.(-2)^2)[[2]]
>>> [1] -2
>>>> mode(bquote(.(-2)^2)[[2]])
>>> [1] "numeric"
>>>
>>> The difficulty is that the tokenizer, which recognizes language elements before the parser goes to work on the grammatical structure, is unable to distinguish the -2 in "-2 + 2" from the one in "-2 ^ y".
>>
>> Are you saying that the tokenizer could be made smarter and recognize
>> "-2" as a token in "-2 ^ y", just because you didn't put a space between
>> "-" and "2"? So in "-2 ^ y" it would be token, but not in "- 2^y".
>> An therefore precedence would now depend on whether there is a space
>> after the minus or not, or something like that? I'm confused.
>>
>
> Nope, precedence would not change at all. "-2" and "- 2" are two different things - in case you didn't realize that spaces are important, try "a<-2" vs "a< -2".

Yes, there is an precedent where space matters. I just hope you're not
saying you'd like to see that repeated for "-2" vs "- 2".

> So, again, if -2 is a value then there is no precedence issue, since you have only operator in  "-2 ^ y" namely "^".

If you make -2 a value, then yes you *effectively* change precedence
from a user point of view. You can be picky about terminology and argue
that it's not an "operator precedence" issue because there is no
operator involved but that doesn't change the fact that now it's a very
bad idea. I can't believe serious programmers are discussing ways to
make -2^2 return 4.

H.

> In "- 2 ^ y" you have two operators, so precedence matters.
>
> Cheers,
> Simon
>
>
>
>> H.
>>
>>> And since constants are generated by the tokenizer, negative ones are not generated. I don't think it is completely out of reach for the parser to recognize the pattern "unary minus a numeric constant" and fold it into a constant of the opposite sign, but I'm not volunteering... (and anyways, it is part of the bigger issue of general constant folding --- I suppose that Luke has a handle on that.)
>>>
>>>
>>>>
>>>>> On 05/07/2014 09:16 AM, John Chambers wrote:
>>>>>> On 5/7/14, 5:21 AM, Therneau, Terry M., Ph.D. wrote:
>>>>>>> Hadley asked about the Blue book; my shelf still has the earlier brown
>>>>>>> book
>>>>>>>    Becker and Chambers, 1984, S: An interactive environment for data
>>>>>>> analysis and graphics.
>>>>>>
>>>>>> Historically interesting, but there was never a guarantee that Version 3
>>>>>> of S (the "blue book") was back-compatible with earlier versions.  We
>>>>>> gave users some help in "getting on the road" to converting, that was
>>>>>> all (see Appendix 4 to the blue book).
>>>>>>
>>>>>> For that one brief moment, we felt free to innovate.
>>>>>>
>>>>>> John
>>>>>>
>>>>>>>
>>>>>>> The manual page for precedence is
>>>>>>>
>>>>>>> $               component select
>>>>>>> %x              special operator
>>>>>>> -               unary minus
>>>>>>> :               sequence operator
>>>>>>> ^ **            exponentiation
>>>>>>> * /             mult/div
>>>>>>> + -             add/sub
>>>>>>> < > <= >= == != logical
>>>>>>> !               not
>>>>>>> & |             and/or
>>>>>>> <- ->           assignment
>>>>>>>
>>>>>>> Terry Therneau
>>>>>>>
>>>>>>> ______________________________________________
>>>>>>> R-devel at r-project.org mailing list
>>>>>>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>>>>>
>>>>>> ______________________________________________
>>>>>> R-devel at r-project.org mailing list
>>>>>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>>>>
>>>>> --
>>>>> Hervé Pagès
>>>>>
>>>>> Program in Computational Biology
>>>>> Division of Public Health Sciences
>>>>> Fred Hutchinson Cancer Research Center
>>>>> 1100 Fairview Ave. N, M1-B514
>>>>> P.O. Box 19024
>>>>> Seattle, WA 98109-1024
>>>>>
>>>>> E-mail: hpages at fhcrc.org
>>>>> Phone:  (206) 667-5791
>>>>> Fax:    (206) 667-1319
>>>>>
>>>>> ______________________________________________
>>>>> R-devel at r-project.org mailing list
>>>>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>>>>
>>>>
>>>> ______________________________________________
>>>> R-devel at r-project.org mailing list
>>>> https://stat.ethz.ch/mailman/listinfo/r-devel
>>>
>>
>> --
>> Hervé Pagès
>>
>> Program in Computational Biology
>> Division of Public Health Sciences
>> Fred Hutchinson Cancer Research Center
>> 1100 Fairview Ave. N, M1-B514
>> P.O. Box 19024
>> Seattle, WA 98109-1024
>>
>> E-mail: hpages at fhcrc.org
>> Phone:  (206) 667-5791
>> Fax:    (206) 667-1319
>>
>

-- 
Hervé Pagès

Program in Computational Biology
Division of Public Health Sciences
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N, M1-B514
P.O. Box 19024
Seattle, WA 98109-1024

E-mail: hpages at fhcrc.org
Phone:  (206) 667-5791
Fax:    (206) 667-1319