Testing by @tclune found a weird failure in pfio unit tests in MAPL (see https://github.com/GEOS-ESM/MAPL/issues/2641). Further testing found the bug could be triggered in a standalong netCDF-Fortran code.
Eventually, it was found that setting either -init=snan,arrays or -fpe0 caused the nf90_create call to crash.
This PR removes those flags if the ifort version is 2021.10 or higher.
NOTE: I have no idea if this is zero-diff or non-zero-diff. I'm not sure if -fpe0 causes a change...
Testing by @tclune found a weird failure in pfio unit tests in MAPL (see https://github.com/GEOS-ESM/MAPL/issues/2641). Further testing found the bug could be triggered in a standalong netCDF-Fortran code.
Eventually, it was found that setting either
-init=snan,arrays
or-fpe0
caused thenf90_create
call to crash.This PR removes those flags if the
ifort
version is 2021.10 or higher.NOTE: I have no idea if this is zero-diff or non-zero-diff. I'm not sure if
-fpe0
causes a change...