Add reactions variant to support VANTAGE-Reactions coupling.

[mrhardman]

To be able to compile VANTAGE-Reactions with Hermes-3, we need to make these modifications to the package definitions. See, e.g., UKAEA-Edge-Code/hermes-3#1 and UKAEA-Edge-Code/hermes-3#3 for context.
We add

• A +reactions variant, so named because that is the package name for VANTAGE-Reactions.
• A CMake compiler flag option to be compatible with Make VANTAGE-Reactions and NESO-Particles features compile only if -… UKAEA-Edge-Code/hermes-3#3.

[oparry-ukaea]

I think the only question here is whether "reactions" is the right name for the variant.

Admittedly "vantage" might not mean much to somebody spack-installing hermes for the first time, but maybe that's ok if we document it properly? Would "kinetic_neutrals" make sense?

[mrhardman]

I think the only question here is whether "reactions" is the right name for the variant.

Admittedly "vantage" might not mean much to somebody spack-installing hermes for the first time, but maybe that's ok if we document it properly? Would "kinetic_neutrals" make sense?

At the moment all this variant does is pull in VANTAGE-Reactions for some test compilation and for use of NESO-Particles in the proxy app. I picked this name just because that is what the package *.py file is called within VANTAGE-Reactions. If there is a more objective name, I am happy to change it.

[oparry-ukaea]

Yeah, fair enough.
@mikekryjak - any preference for the name of a spack variant that turns on VANTAGE-Reactions features in Hermes?

[mikekryjak]

Sorry @oparry-ukaea @mrhardman, I totally missed this.
While I think it doesn't really matter what the name is at the moment as things are still in flux, I would prefer the "kinetic neutral variant" of Hermes-3 to be called "VANTAGE" not "reactions" because this way the name would better describe what capability is being enabled.

[mrhardman]

For the variant name, would vantagereactions be acceptable? @mikekryjak @oparry-ukaea

[will-saunders-ukaea]

How does this spack repo get configured to use the NESO and VANTAGE spack repos? It seems that if it depends on those repos existing in submodule pointers in BOUT then this BOUT-Spack repo is not self contained and can only be used as a submodule of BOUT (at which point does it make sense to be a submodule)?

[oparry-ukaea]

Thanks Will, I hadn't clocked that having spack clone the VANTAGE-Reactions repo as a git submodule wouldn't, on its own, allow it to use the packages therein...

@mrhardman Will has pointed me to a possible solution - I'll discuss with you offline before we update this issue.

[oparry-ukaea]

Unfortunately the solution that we had in mind isn't going to work - loading packages from a git url works in a repos.yaml (plural) for a particular environment, but can't, as far as I can see, be used to tell one package repository about another via its repo.yaml (singular).
I've made a PR to warn anybody trying to use the variant without the reactions package being defined.

[oparry-ukaea]

@mrhardman If you're happy with #12, I think we just need to merge that, update the variant name to 'vantagereactions' and then this can go in.

[mrhardman]

@mrhardman If you're happy with #12, I think we just need to merge that, update the variant name to 'vantagereactions' and then this can go in.

Sounds good to me.