Installation issue: `seacas 2022-10-14:2023-05-30` with `fmt@9` when `gcc` mixed with `apple-clang` about spack

[gsjaardema]

I am fine with the suggested change. Please feel free to submit a PR if you ahve time. Thanks for taking the time to look into this.

[Chrismarsh]

The above linked fmt git issue suggests removing FMT_ENABLE_IF to solve it. However my patches around that have been unsuccessful.

Because it is the mismatch between seacas%apple-clang and fmt%gcc, my initial reaction is to conflict on ^fmt%gcc. Would this be an ok addition to seacas?

[haampie]

Hm, maybe Spack got a bit too greedy with reuse, and it needs a more generic fix.

[gsjaardema]

Is this specific to seacas, or would any client of fmt have the same issue when using different compilers? In other words, does the seacas package.py need a fix, or does the fmt package.py need a fix?

[Chrismarsh]

In this case, this is a specific dependency of seacans that in a mixed toolchain will reliably fail. I think that seacans package.py should protect against this case.

More broadly, I think this absolutely happens throughout spack when mixing the two -- I might be seeing this in vtk but I need to investigate more.