Message-ID: <19931.56073.758992.208514@max.nulle.part>
Date: 2011-05-24T16:21:29Z
From: Dirk Eddelbuettel
Subject: [Rcpp-devel] Rcpp::wrap seg fault (when called with a NULL pointer)
In-Reply-To: <4DDBCED7.5040909@decsai.ugr.es>
On 24 May 2011 at 17:29, Christoph Bergmeir wrote:
| Hi,
|
| of course it's your decision, but I personally think that adding the check to the wrapper is a good idea. For now, I added the check everywhere in my code where I use a const char* with Rcpp::wrap.
Please try SVN revision 3030 which is now resilient to this particular case
of NULL:
edd at max:~$ r -p /tmp/nil.r
Loading required package: inline
Loading required package: methods
NULL
Bye
edd at max:~$ cat /tmp/nil.r
require(inline)
fun <- cxxfunction(signature(),plugin="Rcpp",body='
const char *ret = NULL;
return Rcpp::wrap(ret);
');
fun()
cat("Bye\n")
edd at max:~$
As I joked earlier, it won't help the next time NULL comes hiding as the
payload of another pointer, but at least const char* is now covered. I also
added two quick unit tests.
Thanks for alerting us to this. Segfaults are after all almost always bad.
Dirk
--
Gauss once played himself in a zero-sum game and won $50.
-- #11 at http://www.gaussfacts.com