[Rcpp-devel] Rcpp::wrap seg fault (when called with a NULL pointer)

[Dirk Eddelbuettel]

On 24 May 2011 at 14:53, Romain Francois wrote:
| Le 24/05/11 13:46, Dirk Eddelbuettel a écrit :
| >
| > On 24 May 2011 at 13:22, Christoph Bergmeir wrote:
| > | Hello everybody,
| > |
| > | I just ran into the following problem: Using Rcpp::wrap with a NULL pointer crashes my R. I'm not so sure if this is a bug or if I'm using Rcpp::wrap in a way it shouldn't be used, so I wanted to discuss this here.
| > |
| > | Here is some example code:
| > |
| > | This works fine:
| > | #-----------------------------------
| > | library(inline)
| > |
| > | src<- '
| > |      const char* ret = "test";
| > |      return Rcpp::wrap(ret);
| > |      '
| > |
| > | fun<- cxxfunction(signature(), src, plugin="Rcpp")
| > | fun()
| > | #-----------------------------------
| > |
| > | Whereas this crashes R:
| > | #-----------------------------------
| > |
| > | src<- '
| > | const char* ret = NULL;
| > | return Rcpp::wrap(ret);
| > | '
| > |
| > | fun<- cxxfunction(signature(), src, plugin="Rcpp")
| > | fun()
| > | #-----------------------------------
| >
| > I think this is mostly of the
| >
| >      Q  Doctor, doctor it hurts when I do this!
| >      A  Well then don't it
| >
| > variety.  wrap() is tasked with making a valid SEXP out of what you hand it.
| > Maybe you want to test for NULL before calling it.
| >
| > Dirk
| 
| And you start sounding like Steve Jobs :-)
| Perhaps we can bring the test inside Rcpp in this wrap :
| 
| inline SEXP wrap(const char* const v ){
| 	return Rf_mkString(v) ;
| }
| 
| Not sure what the result ought to be though: NULL, NA, "" ?

Yes, when glancing at the traits header for wrap, and the internal/wrap.h
which contains this I also spotted this special case for const char*, and had
the same idea.  This can hold a test for NULL.

But that said I simply fear that by accomodating const char* we set ourselves
up for the next guys sending us a void*   As the saying goes, you cannot
write foolproof software as you simply attract smarter fools ;-)

We could indeed return R_NilValue in this case.  But what about other 'misuses'?

Dirk

| 
| Romain
| 
| > | I'm using Rcpp 0.9.4 and R.2.13, which I think are the latest versions.
| > |
| > | Regards,
| > | Christoph
| > |
| > | PD: Dirk, Romain, I just read through the slides you made for this workshop a month ago. Great work! Thanks a lot!
| >
| > Thanks!
| >
| > |
| > |
| > | --
| > | Christoph Bergmeir
| > | e-mail: c.bergmeir at decsai.ugr.es
| > | Grupo SCI2S, DiCITS Lab          (http://sci2s.ugr.es/DiCITS)
| > | Dpto. de Ciencias de la Computacion e Inteligencia Artificial
| > | E.T.S. Ingenierias de Informatica y Telecomunicacion
| > | Universidad de Granada
| > | 18071 - GRANADA (Spain)
| > | _______________________________________________
| > | Rcpp-devel mailing list
| > | Rcpp-devel at lists.r-forge.r-project.org
| > | https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
| >
| 
| 
| -- 
| Romain Francois
| Professional R Enthusiast
| +33(0) 6 28 91 30 30
| http://romainfrancois.blog.free.fr
| http://romain-francois.com
| |- http://bit.ly/hdKhCy : Rcpp article in JSS
| |- http://bit.ly/elZJRJ : Montpellier Comedie Club - Avril 2011
| `- http://bit.ly/fhqbRC : Rcpp workshop in Chicago on April 28th
| 
| 

-- 
Gauss once played himself in a zero-sum game and won $50.
                      -- #11 at http://www.gaussfacts.com