On 10 May 2019 at 18:31, Kurt Hornik wrote: | Thanks. Btw, Tomas just told R Core that | | ********************************************************************* | I had a quick look at reference lapack builds in different | distributions: looking at the disassembly, and specifically for dposv | tail-calling into dpotrs. I checked the latest packages from Fedora 30, | Fedora Rawhide (the same?), Ubuntu 19.04, Debian Sid, OpenSuse Leap | 42.3. None of the builds had that problem (yet). I've been looking for | the packages via pkgs.org. | ********************************************************************* | | so we should be safe re BLAS/LAPACK binary packages in testing for now. | (But of course, this might change in case of rebuilds using current | gfortran-8).
Well as we said for some sane distributions (us ;-) gcc-9 is still of limits, and the patch to gfortran-8 is AFAIK not that old. Then again the analysis is also incomplete as some distributions (us ;-) permit live switching of BLAS/LAPACK so we'd also have to look at disassembly for OpenBLAS, Atlas, ... and whatever else is used. Anyway, it is great to have the extra flag. I just prepared r-base 3.6.0-2 and it contains the new flag in /etc/R/Makeconf thanks to your changes to configure{,.ac} which I "backported" to 3.6.0-2. It also includes the whitelist for the non-portable compiler flags noise. Dirk -- http://dirk.eddelbuettel.com | @eddelbuettel | e...@debian.org _______________________________________________ R-SIG-Debian mailing list R-SIG-Debian@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-debian