[Rcpp-devel] Rcpp 0.10

[terrance savitsky]

Thank you JJ and Dirk,  For whatever it's worth, I had already pulled the
attributes vignette and read it multiple times  before making a posting. I
found the vignette to be very clear until the section on working under
packages.  As mentioned, I had some difficulty to discern the old-to-new
comparison of developing a package with Rcpp.  I found a note in this
posting thread that asked similar questions to what i was thinking.  the
answer noted that the new features wouldn't preclude the old, but didn't
comment on developing packages; hence, my note.   I appreciate the coaching
and will build a package skeleton to see what i may learn.  big picture, on
future packages I compose under use of Rcpp::export and inclusion of
roxygen2-processed documentation at the c++ layer, i will no longer plan to
create an R function wrapper for .Call(c++ functions)  at the R layer of my
package.

Terrance.


On Wed, Nov 21, 2012 at 7:10 PM, JJ Allaire <jj.allaire at gmail.com> wrote:

>
> All to say, it is unclear to me whether the intended use of
>> compileAttributes() is solely to help package writers allow package users
>> to more easily access the c++ functions at the R layer or additionally to
>> also streamline how package developers source c++ files within their own
>> package development process under Rcpp.
>
>
> It is indeed intended to streamline how package developers work. With
> compileAttributes you only need to create cpp files that contain function
> definitions marked with Rcpp::export attributes (you don't need to write
> any explicit code in .h or .R files, this code is generated by
> compileAttributes).
>
> If you pull the latest Rpp from R-forge  you'll find:
>
> (a) An expanded vignette which goes into more details about the whole
> mechanism (especially as it relates to packages).
>
> (b) An updated version of Rcpp.package.skeleton which you can optionally
> pass a parameter of attributes = TRUE to see what hello, world using
> attributes looks like.
>
>



-- 
Thank you, Terrance Savitsky
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/rcpp-devel/attachments/20121122/1d2a5869/attachment.html>