Merge into upstream wasm-tools?

[alexcrichton]

I wanted to open an issue on this to perhaps discuss this. Currently wasm-tools compose prints a warning (rightfully) to use wac and in the future the plan is to sunset wasm-tools compose and the wasm-compose crate entirely. That being said one of the key benefits of wasm-tools compose is that it's "always in sync" with the rest of wasm-tools, basically entirely because it's in the same repository. I like the idea of having the two in sync, so wanted to raise the question: would it makes sense to migrate this repository to living within wasm-tools?

The benefits of such a merge could be:

• Integrated release cadence with wasm-tools
• Up-to-date bindings to wasm-tools crates
• Up-to-date on the latest component model features
• (maybe) Better visibility for functionality

Downsides, however, are:

• Monorepo vs multi-repo - CI is more complicated, sweeping changes to the monorepo are harder (e.g. not all contributors will be familiar with wac)
• Less focused issue tracker (more monorepo vs multi-repo)
• More frequent "breaking" releases for functionality that's probably not actually breaking

Would maintainers/folks here be interested in such a merge? Or best to keep things separate?

[alexcrichton]

Another possibility: merging some of the core functionality such as document parsing, resolve, etc, into wasm-tools but otherwise leaving out anything that would require networking. That would mean that the wac CLI would still exist for the purpose of doing networking and having various defaults, but for purely local cases wasm-tools compose could be repurposed to have simlar functionality to wac (and be backed by wac)

[fibonacci1729]

Personally I'm on board with this 👍 I like the plan to keep the cli separate so as to not introduce registry networking shenanigans into wasm-tools and just take a dependency here on wasm-tools

[fibonacci1729]

I'm going to start digging into this this week.

First going to bring the various crates up-to-date with wasm-tools.

Just saw #154 to bring things mostly up-to-date.