On 19 January 2021 at 17:45, Florian Rupprecht wrote:
| Hello,
|
| the CRAN checks show an error for my newly released package 'httpgd' with
| 'r-patched-solaris-x86':
|
| Error: C++17 standard requested but CXX17 is not defined
|
| It seems like 'rcppsimdjson' had the same problem (
| https://github.com/eddelbuettel/rcppsimdjson/issues/7) and worked around it
| by disabling some functionality with compiler flags.
Clarification: An earlier release required C++17. Later releases relaxed
that. We only need C++11 these days. (And I just noticed that I should
updated Makevars.win accordingly but it really isn't a constraint with
Rtools4).
The test is still in configure though, and it appears that Solaris is still
having the same issue. Here is what I did (where configure is once again a
shell script):
CXX17=`${R_HOME}/bin/R CMD config --all | awk '{print $1}' | grep "^CXX17$"`
[...]
if test "x${CXX17}" = "x"; then
echo "** this is not a system on which RcppSimdJson can be built"
echo "** setting C++11 as a fallback"
sed -e 's/@CXXSTD@/CXX11/' src/Makevars.in > src/Makevars
else
| I don't really know what the error means for the CRAN package. Will it be
| removed if it is not fixed? The R extension documentation states that C++17
| is supported since R 3.4.0, and I do have R >= 4.0.0 as well as C++17
| specified in the DESCRIPTION file.
| Or is there some way to indicate that Solaris is not supported? (I doubt I
| will ever have users running it.)
Fair question. I don't know the answer. As package authors we have no
sanctioned way to opt out. FWIW I think the most 'aggressive' my packages
are in terms of C++ standard is currently C++14 (as e.g. RcppGeiger, which
isn't even on CRAN).
httpgd is a cool package by the way, I gave plotting to the webbrowser a
quick spin this morning.
Dirk
--
https://dirk.eddelbuettel.com | @eddelbuettel | e...@debian.org
______________________________________________
R-package-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-package-devel
