[Rcpp-devel] Package Submission Barred by Pedantic

Dirk Eddelbuettel edd at debian.org
Mon Dec 17 20:53:48 CET 2018


On 17 December 2018 at 13:30, Nicholas Rittler wrote:
| I've had no luck arguing that this warning is a false positive with Uwe
| Ligges. How long does this usually take?

Nobody knows. If in doubt, try to make the issue go away.
 
| What is more, while the issue has been resolved in the Boost code, it seems
| to my untrained eye that winbuidler is indeed setting "-pedantic", which
| it's not supposed to do. Is that not the implication of the following lines
| from my winbuilder logs, along with the fact that I don't set -pedantic in
| my Makevars.win or Makeconf file?:
| 
| *** arch - x64
| d:/Compiler/gcc-4.9.3/mingw_64/bin/g++ -m64
| -I"D:/RCompile/recent/R/include" -DNDEBUG -I.
| -I"d:/RCompile/CRANpkg/lib/3.6/BH/include"
| -I"d:/RCompile/CRANpkg/lib/3.6/bigmemory/include"
| -I"d:/RCompile/CRANpkg/lib/3.6/Rcpp/include"
| -I"d:/RCompile/CRANpkg/lib/3.6/RcppArmadillo/include"
| -I"d:/Compiler/gcc-4.9.3/local330/include"     -pedantic -O2 -Wall
| -mtune=core2 -c RcppExports.cpp -o RcppExports.o

You can never delete flags for your _package_ which are coming from the _R
binary_ itself. C'est la vie.

Can you remind us _what your error was_ ?  There are a bazillion config
options to Boost.  Eg adding -DBOOST_NO_AUTO_PTR tells the Boost library no
to use an idiom about which newer compiler complain a lot. (And eg I sent a
PR to the dplyr team to add this which they, after sitting on it for three
weeks, now merged.)

| On Mon, Nov 5, 2018 at 11:34 PM Qiang Kou <qkou at qkou.info> wrote:
| 
| > Hi, Nicholas,
| >
| > First, what you reported is a known issue from boost [1], not Rcpp.
| >
| > Second, this issue has been fixed in boost 1.69 [2] and the bh package

What is the issue again? I have BH 1.69 ready but cannot upload it as three
packages regress so need to make time to write to their maintainers.

Dirk

-- 
http://dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org


More information about the Rcpp-devel mailing list