[Rcpp-devel] Rcpp and required steps before next release

Dirk Eddelbuettel edd at debian.org
Tue Jan 28 17:35:56 CET 2014

Hi Steffen,

Thanks so much for the prompt response.  Replying to both posts here.

On 28 January 2014 at 17:08, Steffen Neumann wrote:
| Hi  Kevin and Dirk,
| thanks for your preparations to make the transition smooth.
| On Mon, 2014-01-27 at 19:35 -0600, Dirk Eddelbuettel wrote:
| [...]
| > linking step, and src/Makevars will go away for most packages. [ Exceptions:
| I pushed Rdisop 1.23.2 to BioC-SVN, removing Rcpp from Depends, 
| as it already depends on RcppClassic. R CMD check's fine,
| with Rcpp and RcppClassic from SVN:
| R version 3.0.2 (2013-09-25)
| [...]
| other attached packages:
| [1] Rdisop_1.23.2       RcppClassic_0.9.4.1
| loaded via a namespace (and not attached):
| [1] Rcpp_0.10.6.1
| Was that all for Rdisop ? 

Seemingly. If it tests fine...

| Do I need to cleanup the Makevars ?

When you release today, you release against Rcpp 0.10.6 on CRAN. The patches
accomodate a transition to make our next upload possible.  

You can only depend on Rcpp 0.11.0 once it out -- which we are trying to make

| Do I need a versioned Depends: for RcppClassic now ?

I don't think so.  

On 28 January 2014 at 17:24, Steffen Neumann wrote:
| I just pushed mzR-1.9.4 to BioC-SVN, which only required a change 
| of #include's in src/pwiz/utility/misc/sha1calc.cpp
| "[...] because it tries to include Rcpp/iostream/Rostream.h, 
|  which no longer exists 
|  (consolidated into Rcpp/iostream/Rstreambuf.h) [...]"
| => Since when does Rstreambuf.h exist ? (Almost) always ? 
|    Or should we have a versioned Dependency on Rcpp here ?

You should include   Rcpp.h   as everything just works.

You only need versioned depends if your code has an explicit dependency on a
feature introduced at a given point --- ie once you simplify src/Makevars you
will need Rcpp (>= 0.11.0).   Makes sense?

And FWIW my write-up didn't cover BioC though. 

And I am at this point less concerned with BioC -- as we have to get onto
CRAN first.  We did of course test against BioC (thanks to Kevin) to see
another code base, and we want to make that smooth.  But let's maybe deal
with CRAN first...

Dirk Eddelbuettel | edd at debian.org | http://dirk.eddelbuettel.com

More information about the Rcpp-devel mailing list