We’ve just posted a roadmap for Apollo Server 4! We’re really excited about the plans in it, which mostly build on a proposal which is the second-most-upvoted issue in the repo.
At a high level, we want to move away from a world where the core team needs to personally improve (or at least evaluation PRs to improve) the integration between Apollo Server and 9 (!) different web frameworks and serverless environments. We believe that these integrations are best maintained by people who use them every day. The main goal of Apollo Server 4 will be to make building an integration between Apollo Server and a web framework a straightforward well-documented process, so that the core team can focus on adding overall platform features and actual users of integrations can focus on improving the integrations.
As part of this, we will be migrating the Fastify, Hapi, Koa, Micro, Lambda, Google Cloud Functions, Azure Functions, and Cloudflare Workers integrations from core-maintained to community-maintained. And that means that we’re looking for folks interested in being those community maintainers!
We’ve opened an issue for each integration where folks can talk to each other and us about their interests and ideas for the future of these integrations. In some cases this is an exciting opportunity to experiment: for example, there are several possible approaches to building a Lambda integration (the current Express-based approach? the Middy middleware framework? etc), which could all be done in different packages.
We do still care about the ability of Apollo Server users to be successful in environments other than Express (heck, we use it with Google Cloud Functions ourselves) and will be happy to provide advice and some contributions when appropriate. But practically speaking, having these packages in core has been more of a hindrance to their users than a help. I look forward to a day where PRs for apollo-server-hapi are reviewed by people who don’t literally have to read the Hapi quickstart guide every time they consider merging one (I’m talking me here).