[Rcpp-devel] Interrupts in compiled code.

[Andrew Redd]

Thanks for pointing that out.  I don't pretend to understand the reasons but
that method seems to be much more responsive than the throw/catch method.
 It seems like the throw catch should be just as efficient, and should add
very little overhead.  It does not add overhead, but the R_CheckUserInterupt
also does not appear to add significant overhead either.  The throw method
seems to take a very long time between when the interrupt is sent before the
error is caught.   Is there a way to speed up the throw/catch?

-Andrew



On Wed, Dec 8, 2010 at 3:42 PM, Davor Cubranic <cubranic at stat.ubc.ca> wrote:

> How about just using R_CheckUserInterrupt()?
>
> Davor
>
>
> On 2010-12-08, at 1:02 PM, Dirk Eddelbuettel wrote:
>
> >
> > On 8 December 2010 at 13:41, Andrew Redd wrote:
> > | I have an MCMC chain that runs entirely in c++ with Rcpp.  It sometimes
> runs
> > | for a vary long time and I want to interrupt it.  Is there an efficient
> easy
> > | way to include catching an interupt signal and either aborting or
> returning
> > | results to that point?  This might be understood well, if so I
> apologize.
> >
> > Great question.
> >
> > Here is a simple example that uses a standard C interrupt handler to set
> a
> > standard C++ exception to abort:
> >
> > ----------------------------------------------------------------
> >
> > #include <csignal>
> > #include <Rcpp.h>
> >
> > void intHandler(int dummy=0) {
> >  std::cerr << "In intHandler" << std::endl;
> >  throw std::runtime_error("Interception caught");
> > }
> >
> > RcppExport SEXP foo(void) {
> >
> >  try {
> >
> >    signal(SIGINT, intHandler);
> >    signal(SIGKILL, intHandler);
> >
> >    while (true) {
> >      sleep(1);                        // not doing much
> >    }
> >    return R_NilValue;
> >
> >  } catch(std::exception &ex) {
> >    std::cerr << "In catch of std::exeception" << std::endl;
> >    // here you insert some clenup
> >    forward_exception_to_r(ex);
> >
> >  } catch(...) {
> >    ::Rf_error("c++ exception (unknown reason)");
> >  }
> >
> >  return R_NilValue;
> > }
> >
> > ----------------------------------------------------------------
> >
> > which I then quickly turned into a shared object, load and ran at the
> > command-line [1]
> >
> > edd at max:/tmp/andrew$ PKG_CPPFLAGS=`r -e'Rcpp:::CxxFlags()'` PKG_LIBS=`r
> -e'Rcpp:::LdFlags()'` R CMD SHLIB int2throw.cpp
> > ccache g++ -I/usr/share/R/include
> -I/usr/local/lib/R/site-library/Rcpp/include     -fpic  -g -O3 -Wall -pipe
> -pedantic -Wno-variadic-macros -c int2throw.cpp -o int2throw.o
> > g++ -shared -o int2throw.so int2throw.o
> -L/usr/local/lib/R/site-library/Rcpp/lib -lRcpp
> -Wl,-rpath,/usr/local/lib/R/site-library/Rcpp/lib -L/usr/lib64/R/lib -lR
> > edd at max:/tmp/andrew$ r -e'dyn.load("int2throw.so"); .Call("foo")'
> > ^CIn intHandler
> > In catch of std::exeception
> > Error in cpp_exception(message = "Interception caught", class =
> "std::runtime_error") :
> >  Interception caught
> > Execution halted
> > edd at max:/tmp/andrew$
> > edd at max:/tmp/andrew$
> >
> >
> > After lauching it from the one-line littler (r) expression, I pressed
> Ctrl-C
> > which then got us into the interrupt which threw the exception which lead
> to
> > the end.
> >
> > In Rscript it looks a little more orderly on the unwind:
> >
> > edd at max:/tmp/andrew$ Rscript -e 'dyn.load("int2throw.so"); .Call("foo")'
> >  C-c C-cIn intHandler
> > In catch of std::exeception
> > Error in cpp_exception(message = "Interception caught", class =
> "std::runtime_error") :
> >  Interception caught
> > Calls: .Call -> cpp_exception
> > Execution halted
> > edd at max:/tmp/andrew$
> >
> >
> > Hope this helps,  Dirk
> >
> >
> > [1] Not sure why I didn't use inline today...
> >
> >
> > --
> > Dirk Eddelbuettel | edd at debian.org | http://dirk.eddelbuettel.com
> > _______________________________________________
> > 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
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/rcpp-devel/attachments/20101208/391fed42/attachment.htm>