What could become a CouchDB toolkit in Rust.
Uses literate programming through tango and tries hard to build nice APIs.
See CONTRIBUTING for a contribution guide.
This is currently mostly API experiments. I’m searching for help with many of them.
Find a list of issues here.
Tango is used to provide richer documentation by using literate programming to make writing descriptions the default.
Features, as experimental as they may be, should never be committed without extensive documentation.
This hasn’t been followed through in the past and has been fixed, but there might be spots. These spots shouldn’t happen again.
Decorated results allow a form of result chaining that hides error handling until the very last step.
See yakshav.es/decorating-results/ about how decorated results work.
The project currently consists of the following crates:
lazers-traits: A set of traits without implementations that describe the general interface towards a CouchDB(-like) database. It defines error and result types and interactions for writing and reading documents.
lazers-hyper-client: A client implementing lazers-traits for a remote CouchDB using hyper. Currently also serves as an implementation example for the interface described in lazers-traits.
lazers-changes-stream: An implementation of the CouchDB changes stream protocol. It is generic over any
Read interface, so it depends on no http client.
lazers-replicator: A (currently unfinished) implementation of the CouchDB replication protocol. Important features of the base libraries were missing, but it is still around as a placeholder or as a way to start.
lazers-liblazers: A C interface to the library. Still figuring a good way to do this. This interface is important, see [Long-Term Goals]
The library itself uses plain Serde for serialisation and deserialisation, this means some boilerplate work is required.
Hat tip to https://lobste.rs/u/gsquire for the final name idea.top