Use the modules: and/or cflags: tokens to further specify a suitable
accompanying gcc version to help pacify picky client packages that ask for C++
standards more recent than supported by your system-provided gcc and its
libstdc++.so.
#+end_quote
*NOTE* the docs say **more recent that ... system**, which might indicate why we still see some problems
- by installing intel-oneapi-compilers[-classic]%gcc@toolchain Spack
automatically configures the intel compiler (bin/icx.cfg and similar) in a way
that it depends on the correct gcc. Passing ~-gcc/gxx-name~ or
~-gcc-toolchain~ explicitly is not required. (We did this in the 24a release
because we did not know about the configuration files; adding these flags has
been removed from mpsd-software in Dec. 2024)
** GCC compatibility
- intel classic (~icc~ and ~icpc~) not compatible with gcc>12 (the compiler has been discontinued, so this is unlikely to ever change)
- This has been [[https://gitlab.gwdg.de/mpsd-cs/spack-environments/-/merge_requests/109#note_1181314][verified by Jehferson Mello]] (observed issue: incompatibility with C++ std library headers)
...
...
@@ -152,7 +151,7 @@ the other packages are compiled
- anaconda3 cannot be used as of 2024 due to license restrictions (MPSD would need to pay for a commercial license)
- we instead provide miniforge3
- we restrict it to the ~conda-forge~ channel (in a ~.condarc~ in the miniforge3 directory)
(the more flexible ~denylist_channels~ introduced in conda 24.9 seems to have a bug)
(the more flexible ~denylist_channels~ introduced in conda 24.9 have a bug)
- users are advised to use the conda module when they want to create their own conda environments
(that way we can ensure no one uses the commercial channels on the HPC system)
- we provide a custom environment, currently called ~python-3.12~ with commonly used packages preinstalled