Several unrelated updates and a rather important bug fix that bring us one step closer to compiling the entire unified environment with the mixed Intel oneAPI compilers (icx, icpx, ifort):
fckit: Straightforward, Upstream PR for spack develop: https://github.com/spack/spack/pull/45196 (includes an earlier update we made in spack-stack-dev for compiling with nvhpc).
qt: First, pulled in updates (two mostrecent commits) for the package from spack develop, then added the workaround for oneAPI. Tested successfully on Blackpearl (compiled qt and ecflow with oneAPI, then start ecflow server and ecflow GUI). Upstream PR for spack develop: https://github.com/spack/spack/pull/45195
Bug fix in lib/spack/env/cc: I mistakenly copied the C++logic into the Fortran block and nobody noticed it. This fixes it. Note that the entire vcheck_flags logic in lib/spack/env/cc will eventually be replaced by a better solution. But since I am currently iterating with the spack developers on how that solution should look like (see https://github.com/spack/spack/pull/44588), let's merge this important bug fix now.
Upstream spack PRs are all approved and in the auto-merge queue - will merge this, because also the parent spack-stack PR of this PR has been approved.
Description
Several unrelated updates and a rather important bug fix that bring us one step closer to compiling the entire unified environment with the mixed Intel oneAPI compilers (icx, icpx, ifort):
fckit
: Straightforward, Upstream PR for spack develop: https://github.com/spack/spack/pull/45196 (includes an earlier update we made in spack-stack-dev for compiling withnvhpc
).qt
: First, pulled in updates (two mostrecent commits) for the package from spack develop, then added the workaround for oneAPI. Tested successfully on Blackpearl (compiledqt
andecflow
with oneAPI, then start ecflow server and ecflow GUI). Upstream PR for spack develop: https://github.com/spack/spack/pull/45195lib/spack/env/cc
: I mistakenly copied the C++logic into the Fortran block and nobody noticed it. This fixes it. Note that the entirevcheck_flags
logic inlib/spack/env/cc
will eventually be replaced by a better solution. But since I am currently iterating with the spack developers on how that solution should look like (see https://github.com/spack/spack/pull/44588), let's merge this important bug fix now.Corresponding spack-stack PR: https://github.com/JCSDA/spack-stack/pull/1195