Type piracy in visualization functions

[Evan Patterson]

The visualization module has lots of type piracies. For example, the Graph constructor is defined for Multispan, Subobject, and ACSetTransformation, but AlgebraicPetri doesn't own any of these types, nor are the methods defined here general enough for those types.

[James]

All of those types have type parameters for the underlying category right? So we could use the type parameters to narrow them to Multispans of ACSetTransformations of Reaction Nets, right? Would that be sufficient to avoid Type Piracy?

[Evan Patterson]

Right, the three cases I mentioned can be fixed by specializing the types in the methods.

There's also stuff like this:

==(a::NodeID, b::NodeID) = isequal(a.name, b.name) && isequal(a.port, b.port) && isequal(a.anchor, b.anchor) && true
hash(a::NodeID, h::UInt) = hash(a.anchor, hash(a.port, hash(a.name, hash(:NodeID, h))))
==(a::Edge, b::Edge) = isequal(a.path, b.path) && isequal(a.attrs, b.attrs) && true
hash(a::Edge, h::UInt) = hash(a.path, hash(a.attrs, hash(:Edge, h)))

If it's helpful to have equality and hash methods for these types, they should be upstreamed to Catlab.

[Andrew Baas]

Great, thanks for catching this! I'll work on getting those types correctly parameterized once we get #69 merged.

It would be helpful to have equality and hash methods for NodeID and Edge, so I'll get an Issue posted and PR made for that.

[Micah Halter]

I think this has been resolved in the latest version of AlgebraicPetri. Let me know if I'm mistaken and I can reopen this!