Rename traits to improve semantics

[mxgrey]

I haven't been very satisfied with some of the trait naming, especially around Buffered and Joined. The -ed suggests that the operation has been performed, so I think Joined makes more sense to describe the value that emerges from the join operation rather than the buffers that are doing the joining. Instead the name for the traits that get applied to collections of buffers have been renamed to use -ing, e.g. Buffering to indicate that they are carrying out the action rather than that the action has been completed (-ed).

I've made the following renames:

• Bufffered -> Buffering
• Joined -> Joining
• Accessed -> Accessing
• JoinedValue -> Joined
• BufferKeyMap -> Accessor

I don't mind iterating on these names. The BufferKeyMap -> Accessor change is one that I especially feel uncertain about. I don't know if Accessor is a clear enough name for the trait, whereas BufferKeyMap is a lot more explicit about its intention.

[koonpeng]

I don't really mind the naming but just to point out that I haven't seen the -ing form (present continuous?) used to name traits before.

[mxgrey]

I haven't seen the -ing form (present continuous?) used to name traits before.

Yeah, it's probably not super conventional, but I think it makes the most sense grammatically for our particular case since we need to distinguish between Joined (an object which has been joined, the output of a join operation) and Joining (objects which are responsible for joining data together). We could consider Joiner instead of Joining but then that would be confusing next to Accessor which is the output of an access operation rather than the thing which performs the access operation (Accessing).

And as far as trait grammar vs convention goes, Clone and Copy should really be Cloneable and Copyable from a grammatical point of view, but they've been simplified to Clone and Copy for the sake of being terse. So I think the conventions are pretty wishy-washy at the end of the day.

Language is hard.