Boundary conditions and labels

[lmoresi]

We currently do not utilise the full functionality of labels in Boundary conditions but we should to save ourselves having to document every deviation we make from PETSc.

[lmoresi]

I'll expand on this: we should adopt the PETSc label approach (many values in a label, pick the ones that are used to provide the constraint). We currently limit this to a single value per label.

[lmoresi]

@julesghub - do you want to task somebody to do this ? It's just because it is a cython pain-in-the-arse to pass lists and arrays from python to petsc that I haven't done this.

The point is that we do not support a common PETSc pattern: to make a single label with multiple values that distinguish different sub-domains. We only support one label value ...

This would save us some effort in documentation but it might make out boundary condition interface more complicated.

[julesghub]

I'm up for looking at this. I think it would be a big benefit for uw3 as we could leverage the PETSc functionality and docs.

[lmoresi]

[ DEMONIC LAUGHTER is heard] "we could leverage the PETSc ... docs"

I think we might need to introduce the idea of a boundary and a sub-boundary or equivalent in order to make this flow within our pattern. It's not too difficult to see how to do this but I think it would be good to do hand-in-hand with a shift in the way we use cython because the painful part of all of this is continually handing python arrays to c arrays.

[lmoresi]

The changes I made are not a fix for this - I still have one label value for one bc. Just letting you know !