I've noticed that we've switched from the old rdf4j-runtime dependency to the storage one on our main "downloads" page for how to get started.
Unless I'm mistaken, this means we no longer package any of the rio stuff. What was the rational for not using runtime anymore?
Not quite. There are two separate "parent" dependencies for people to use:
1. rdf4j-client, which includes all of Rio, as well as the httpclient, repository API and HTTP/SPARQLRepository, and utils, as well as the Model API. Typically what you'd need if you are doing "client-side" RDF processing (parsing, writing, working with in-memory models, talking to a remote endpoint).
2. rdf4j-storage, which includes all of rdf4j-client, as well as the various sail implementations. This is basically the "everything and the kitchen sink".
The only difference with runtime is really that runtime also includes things like the workbench and server. And that is also one of main reasons I am recommending use of rdf4j-storage over rdf4j-runtime: the runtime dependency includes things that are not really relevant to 99% of users (another is that I've always found the name "rdf4j-runtime" awkward and potentially confusing users into thinking that it's a maven runtime-scoped dependency).
Of course, tons of existing projects currently still use rdf4j-runtime as their main dependency, so we will continue to maintain it for the foreseeable future. But we advertise rdf4j-client/rdf4j-storage as the recommended way to import RDF4J into your project.
Cheers,
Jeen
