match="once", match="all_permutations" not described in spec

[lilyminium]

In the virtual sites specification, the behaviour of match is never described in the spec -- only the allowed values. As this affects how many particles are added to the molecule, it's an important aspect of defining vsites in the FF.

The allowed values are also don't fully match the implementation, e.g. it specifies match="once" IF type="TrivalentLonePair" but in the Toolkit match="once" is allowed for DivalentLonePair if outOfPlaneAngle is 0.

[j-wags]

The behavior of the 'name' field and the hierarchy rules for when multiple vsites assigned to the same atom(s) coexist vs overwrite is also unspecified. Currently this behavior only exists as an implementation detail. (This could be a separate issue but I think it's closely related enough that it should go in this one)

[davidlmobley]

I think (?) we discussed this in a meeting and said we should do as Jeff says and just introduce an explanation to the spec. But I'd have to consult our last (?) comittee meeting notes to be sure. I think we just need to make sure the spec explains what is implemented otherwise it (the spec) is not complete.

[mattwthompson]

I bumped into this again today - trying to double-check my understanding against what it's supposed to be. I think I understand the one or two that are ever actually used, but if I was newer I'd probably be guessing