On 6 April 2016 at 18:49, Aurelien Jarno wrote: | On 2016-03-25 09:05, Dirk Eddelbuettel wrote: | > | > On 25 March 2016 at 14:20, Sébastien Villemot wrote: | > | Le mercredi 23 mars 2016 à 15:32 -0500, Dirk Eddelbuettel a écrit : | > | > On 23 March 2016 at 21:08, Matthias Klose wrote: | > | > | On 23.03.2016 20:05, Dirk Eddelbuettel wrote: | > | > | > I just looked at /etc/R/Makeconf where CC, CXX, ... get encoded and they are | > | > | > not versioned (ie NOT gcc-4.9) so I am pretty sure we are not using 4.9 | > | > | > anywhere but on mips. | > | > | | > | > | unfortunately mips is still a release architecture. | > | > | > | > Rock, meet hard place. | > | > | > | > Seb: Any advice? Has this been fixed? Do we need to get Aurelien and the | > | > mips folks involved? | > | | > | As explained by Aurelien Jarno in <[email protected]>, | > | this is indeed a kernel issue. | > | | > | A patch exists, but I don't know if it has been applied on buildds. I | > | guess you should contact Aurelien directly. | > | > Thanks! | > | > Doko: what do you want me to do here? Remove the gcc-4.9 invocation used | > only on mips, and hence loose mips? Won't the package then get tossed from | > all other architectures? | | All the mips build daemons no have a fix. It is therefore fine to remove | the build-depends on gcc-4.9 and build the package with the default GCC | on all architectures.
Will do. In fact, I just updated it in my sources. We are currently creating approximately-weekly builds of beta, then rc, then release tarballs of R leading up to the release of R 3.3.0 which is scheduled for May 3. [1] As I just made a release this morning, this will affect the next build in a few days. Dirk [1] There are transition issues on Windows with an (overdue !!) upgrade of the toolchain to g++-4.9.3 that upstream needs some extra time with. The release was originally planned for next week. -- http://dirk.eddelbuettel.com | @eddelbuettel | [email protected]
