[Rcpp-devel] Rcpp version of DATAPTR

[Whit Armstrong]

Ok, I see.

My package includes some headers which have direct includes of the
armadilo headers (hence, they are grabbing the deb testing version of
Armadillo).

I normally put system headers above R headers, but I think if I
reverse this convention for RcppArmadillo, and let the include guard
kick in, then I should be all set...

and yes.  Looks good.  Compiles.  No test failures.

Thanks again.

-Whit


On Wed, Apr 11, 2012 at 3:54 PM, Dirk Eddelbuettel <edd at debian.org> wrote:
>
> On 11 April 2012 at 15:32, Whit Armstrong wrote:
> | Thanks, Dirk.
> |
> | one more quick q.
> |
> | Should I be doing something to prevent these warnings:
> |
> | /usr/local/lib/R/site-library/RcppArmadillo/include/RcppArmadilloConfig.h:64:0:
> | warning: "ARMA_DEFAULT_OSTREAM" redefined [enabled by default]
> | /usr/local/lib/R/site-library/RcppArmadillo/include/armadillo_bits/config.hpp:73:0:
> | note: this is the location of the previous definition
> |
> | Or is this the intended effect of remapping std::cout to Rout?
>
> That very much looks like either a very outdated version where Conrad's
> definition is not #ifndef protected.  His (in config.hpp) should be
>
> #if !defined(ARMA_DEFAULT_OSTREAM)
>  #define ARMA_DEFAULT_OSTREAM std::cout
> #endif
>
> which it has been for a few releases.
>
> Or you are including header files by force a second time, which we recommend
> not to.  We recommend a single '#include <RcppArmadillo.h>' with which
> everything is fine.   If you do include both, and in the wrong order, then
> you are, as they say, on your own...
>
> Dirk
>
> --
> R/Finance 2012 Conference on May 11 and 12, 2012 at UIC in Chicago, IL
> See agenda, registration details and more at http://www.RinFinance.com