Brennan Saeta

Some pointer sugar.

1

This, unfortunately, is somewhat unsatisfying, as it cannot obviate all need for `.pointee`. In particular, it cannot satisfy the case when the pointee needs to be modified in some fashion....

Refactor co-primes algorithm to be generic & avoid allocation.

7

Thank you @dabrahams for the suggestions!

Implement `UnmanagedBuffer` and switch `TaskDeque` to use it.

1

Previously, `TaskDeque` was implemented in terms of `ManagedBuffer`. While `ManagedBuffer` implements the semantics we'd like, it is implemented as a class. This can induce a significant amount of reference couting...

This PR adds functionality to make it easy to combine the edges of 2 graphs, and the supporting functionality to implement this capability. There are 2 ways to combine the...

Rewrite Deque

3

The previous implementation did not appropriately conform to the standard Collection protocols, and additionally had extra reference counting operations. This new Deque implementation is designed to be more flexible and...

So that these tests can be re-used in dependent packages.

Point users to `vertices.count` and `edges.count` instead, and add to documentation `VertexListGraph` and `EdgeListGraph` that conforming graphs should ensure the `count` properties of `vertices` and `edges` are as efficient as...

Potential names to re-think: - `VertexId` / `EdgeId` / `Edge` / `Vertex` and [associated documentation](https://github.com/saeta/penguin/pull/84/files#r440521494). - `source(_:)` / `destination(_:)` - `VertexListGraph` / `EdgeListGraph` Partial context: https://github.com/saeta/penguin/pull/84/files#r440518969, https://github.com/saeta/penguin/pull/84/files#r440526601