Path search with cost function #24
Conversation
core/src/main/scala/Graph.scala
Outdated
| * Cheapest path from vertex `s` to vertex `t` under the cost function `costFkt` with labels | ||
| * @group bfs | ||
| */ | ||
| def cheapestPath[C](s: N, t: N, costFkt: (LNode[N,A],B,LNode[N,A]) => C)(implicit num: Numeric[C]): Option[LPath[N,B]] = { |
There was a problem hiding this comment.
This could be a little more generic as C: Monoid: Order. I imagine C would usually be numeric in the real world, but it doesn't have to be.
core/src/main/scala/Graph.scala
Outdated
| */ | ||
| def cheapestPath[C](s: N, t: N, costFkt: (LNode[N,A],B,LNode[N,A]) => C)(implicit num: Numeric[C]): Option[LPath[N,B]] = { | ||
| require(contains(s)) | ||
| require(contains(t)) |
There was a problem hiding this comment.
Could we return None here, or is it important to distinguish a missing vertex from an empty searchtree? The exceptions make me a bit uncomfortable.
There was a problem hiding this comment.
Yeah, ideally we don't throw exceptions but instead codify things into the type system :-)
There was a problem hiding this comment.
Rewrite all of this in idris and I will happily encode that in the signature :)
There was a problem hiding this comment.
Well, until someone publishes quividris, my question remains. :) Would a None be an acceptable result, or do we need a third value to return?
There was a problem hiding this comment.
Sorry for not answering this sooner. I think that would be perfectly fine.
- Redecoration includes the focus of the decomposition when acting on the rest of the graph - Fix GDecomp generator so incorrect instances cannot be built (this should probably be a smart constructor)
|
Uh-oh. This doesn't compile together with the changes to #23 and #26 are breaking changes. I'm planning to open a quiver6 branch so we only bump the major version once with our automated release process. We can target that with whatever comes of your open PRs and get a release cut before the cats port. |
|
I can fix this tomorrow. Ideally the LPath stuff would be merged by then. If not, I'll cherry pick it into this branch as well. |
|
Great. I'll merge #23 into the quiver-6 branch now, and you can target that branch with your fixed up version of this, and we'll be ready to release. |
|
Closing because now targeting quiver-6 in #28 |
Functionality for finding the cheapest path between two nodes according to a cost function that sees both nodes and edge labels.