Just to follow up myself, it turns out that it¹s easy to redirect cerr and
cout to files, and that solves all my issues.
In case anyone is interested, here is how I did it:
#include<ostream>
#include<fstream>
RcppExport SEXP myfunction(SEXP params)
{
try
{ // or use BEGIN_RCPP macro
//start off by redirecting
streambuf* save_sbuf_fout;
streambuf* save_sbuf_ferr;
streambuf* save_sbuf_cout;
streambuf* save_sbuf_cerr;
ofstream fout;
ofstream ferr;
fout.open("cout.txt");
ferr.open("cerr.txt");
save_sbuf_cout = cout.rdbuf();
save_sbuf_cerr = cerr.rdbuf();
save_sbuf_fout = fout.rdbuf();
save_sbuf_ferr = ferr.rdbuf();
cout.rdbuf(save_sbuf_fout);
cerr.rdbuf(save_sbuf_ferr);
/*do whatever I normally do and write to cerr and/or cout all I
want*/
//revert to original state
cout.rdbuf(save_sbuf_cout);
cerr.rdbuf(save_sbuf_cerr);
fout.close();
ferr.close();
}
}
On 4/27/11 9:26 PM, "Sean Robert McGuffee" <sean.mcguf...@gmail.com> wrote:
>> >On 11/10/2009 12:23 PM, Dirk Eddelbuettel wrote:
>>> >
>>> > On 10 November 2009 at 09:47, Romain François wrote:
>>> > | Hi,
>>> > |
>>> > | Would it make sense to have an rout and rerr so that we could do :
>>> > |
>>> > | rout<< "bla bla"<< endl ;
>>> > |
>>> > | and
>>> > |
>>> > | rerr<< "bla bla"<< endl ;
>>> > |
>>> > | and this would go into Rprintf and REprintf ? or maybe we can directly
>>> > | redefine cout and cerr
>>> >
>>> > Hm. That never really arose in my use. cout was mostly for debugging.
>>> >
>>> > I think a C++-ish way is to create<< operators (that eg RcppDate and
>>> > RcppDatetime have).
>>> >
>>> > Where you thinking of something other than debugging help?
>>
>> |Not really, but just sort of making sure debugging is consistent with
>> |the recommendation from WRE of using Rprintf instead of printf
>
> Has anyone followed up on this? I¹m wondering if there might be a simple way
> to redirect cerr and cout to something like an rerr and rout? I for one have
> thousands of lines of code that use cerr and cout in libraries that I would
> like to use in R. It seems that they currently are causing a crash at run-time
> when I leave those lines in my packages. I won¹t know for sure until I take
> the time to remove all of them, but who knows how long that might take. I
> think it would probably be faster and easier for me to redirect these streams
> to R somehow. Has that been worked out yet? If so, how? If not, I think maybe
> I can redirect them to some log files or something. Has anyone done that?
> Thanks,
> Sean
_______________________________________________
Rcpp-devel mailing list
Rcpp-devel@lists.r-forge.r-project.org
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
