Description
Currently we have two separate but related concepts: Forklift repositories, and Forklift pallets. This is because initially Forklift pallets could only import packages from Forklift repositories. But we've gradually been turning Forklift pallets into a superset of Forklift repositories: now Forklift pallets can deploy locally-defined packages, and they can define packages and they can be required by other pallets as if they were repositories. Currently, it's possible to define and compose pallets without defining repositories at all. So it would be simpler to describe Forklift if we had only the concept of pallets without also having the concept of repositories, since pallets are now basically a superset of the functionality of repositories.
There may be some implications for how we resolve packages and for how we load packages from the cache (since currently pallets are cached on the filesystem in unmerged form, and we load packages from the cache by searching the filesystem) and for how we bundle packages in pallet bundles and for how we load packages from pallet bundles - I haven't thought through those implications yet.