[R] Modify base R functions in Rprofile.site

[Duncan Murdoch]

On 04/10/2009 6:07 PM, Yihui Xie wrote:
> Thanks a lot, Charlie. What a coincidence -- I'm also working on
> Sweave functions. parse() and deparse() can make the code more tidy
> (they are used in source() and RweaveLatexRuncode()),
> 
>> RweaveLatex
> function ()
> {
>     list(setup = RweaveLatexSetup, runcode = RweaveLatexRuncode,
>         writedoc = RweaveLatexWritedoc, finish = RweaveLatexFinish,
>         checkopts = RweaveLatexOptions)
> }
> <environment: namespace:utils>
> 
> but we can either (1) let R automatically 'tidy up' our code and
> remove all the comments or (2) keep the comments but leave the
> original code untouched.
> 
> I worked out a trick to preserve the comments while tidying up the R
> code, and I want to replace parse() and deparse() with my customized
> functions, in which case Sweave will keep the comments.
> 
> However, my functions will need base::parse() and base::deparse() to
> help me, therefore my real difficulty is, how to let Sweave use my
> functions to parse and deparse R code without really modifying them in
> the base enviroment? I'm using the command 'R CMD Sweave'. 

I think it's really unlikely that you need to modify those two 
functions.  If you leave the source references turned on, the standard 
parse will tell you where all the code is; you can then write it out in 
any way you like, calling the standard deparse on bits and pieces, and 
mixing in the comments as you like.

The disadvantage of replacing parse or deparse is the maintenance:  if 
the standard ones change, you have to decide whether to incorporate 
those changes or not, in every case.  Neither parse nor deparse change a 
lot, but there have still been 2-3 dozen changes in the last year.

Now, if you choose instead to modify the original sources, in most cases 
Subversion will correctly merge the new changes in.  So that's better 
than a wholesale replacement, but still not as good as simply working 
with the output of the standard functions.

 > If there is
 > no neat approach, I'll change my question to: how to write a package,
 > say, Sweave2, that can be run using command line 'R CMD Sweave2 file'?

Currently you can do that by writing Sweave2 as a Perl script and 
installing it in the RHOME/bin directory, but that may change:  we are 
working hard to do away with the need for Perl.

Duncan Murdoch

> 
> Regards,
> Yihui
> --
> Yihui Xie <xieyihui at gmail.com>
> Phone: 515-294-6609 Web: http://yihui.name
> Department of Statistics, Iowa State University
> 3211 Snedecor Hall, Ames, IA
> 
> 
> 
> 2009/10/4 cls59 <chuck at sharpsteen.net>:
>>
>>
>> Yihui Xie wrote:
>>> Hi everyone,
>>>
>>> I want to modify two base R functions 'parse' and 'deparse'
>>> immediately after R has started, so I added some code in the file
>>> 'Rprofile.site' under the 'etc' directory. Here is a simple example:
>>>
>>> parse=function(...){
>>>     base::parse(...)
>>> }
>>>
>>> I'll get an error when I start R as follows:
>>>
>>> Error: cannot change value of locked binding for 'parse'
>>>
>>> Is there a solution for my problem? Please note that I *have to*
>>> override the function names. Thanks very much!
>>>
>>>
>>
>> Hi Yihui,
>>
>> I have had cause to do some some similar things to Sweave functions-- maybe
>> the following will help:
>>
>>  # Find out where the function lives.
>>  env <- as.environment( 'package:base' )
>>
>>  # Crack the binding.
>>  unlockBinding( 'parse', env )
>>
>>  # Replace the function.
>>  assignInNamespace( 'parse', function(...){
>>
>>   # Your function here, or an object that contains it.
>>
>>   }, ns = 'base' )
>>
>>  # Relock the binding.
>>  lockBinding( 'parse', env )
>>
>>
>> Note that the above is *very* dark voodoo-- it may cause unforeseen
>> consequences.
>>
>> Good luck!
>>
>> -Charlie
>>
>> -----
>> Charlie Sharpsteen
>> Undergraduate
>> Environmental Resources Engineering
>> Humboldt State University
> 
> ______________________________________________
> 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.