On Thu, Jan 3, 2013 at 11:28 PM, Dirk Eddelbuettel <e...@debian.org> wrote:
> > On 3 January 2013 at 16:01, Dirk Eddelbuettel wrote: > | > | Hi Alexey, > | > | On 3 January 2013 at 22:13, Alexey Stukalov wrote: > | | Dear Rcpp developers, > | | > | | After an upgrade to Rcpp 0.10.2 I've got an error "pairlist undefined" > during > | | the compilation of my Rcpp-using C++ library > | | (actual gcc 4.7 output is hundreds lines long, so I skip it ) in the > | | constructions like this: > | | > | | Rcpp::DataFrame::create( > | | Rcpp::Named( R_COLUMN_OBJECT, objectIds ), > | | Rcpp::Named( R_COLUMN_COMPONENT_INDEX, compIxs ), > | | Rcpp::Named( R_STRINGS_AS_FACTORS, false ) > | | ) > | | > | | whereas with the older versions (including 0.10) it was ok. > | | I figured that pairlist is defined in Rcpp/grow.h. > | | When I include this header before <Rcpp/DataFrame.h>, the problem goes > away. > > I wasn't looking closely enough, in SVN's Rcpp.h we already have this on > lines 38 and 57: > > #include <Rcpp/grow.h> > [...] > #include <Rcpp/DataFrame.h> > > So exactly what was the problem and what was the purported fix? > > Dirk > > Hi Dirk, thanks for taking time to look into this. Well, I guess if I would use <Rcpp.h>, which basically defines everything, the problem will go away. In my project, however, I'm trying to include only essential headers to reduce compilation time. In this case <Rcpp/DataFrame.h>. If Rcpp's header policy doesn't require that the public headers should include the ones they depend upon, the reported thing is not a bug. Alexey
_______________________________________________ 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
