Propose new list of packages in the monorepo to reduce code split

[spalladino]

We currently have around two dozen packages, and interdependencies between them are hard to track. For instance, we already agreed on merging some of them like l2-block and tx into a single one, but we can go further.

Review the list of packages we have today, make a new proposal, get buy in, and implement it. Things to consider:

• Which packages map to low-level repos (eg bb, circuits.js)
• Which packages have shared code vs entrypoints (consider moving them to separate parent folders, as suggested here)
• Which packages will eventually run on the browser, on node, or both

A modern alternative is bundling everything in 1-2 major packages, and make an aggressive use of namespaces, and rely on tree shaking from consumers for reducing bundle size.

[spalladino]

Going through the package list, after introducing the types package, it seems the split makes sense (or I got used to it!). Still, it could be worthwhile organizing the packages depending on their position in the stack. Consider the following:

Entrypoints

apps folder, contains packages that are starting points for different parts of the system (node, rpc-server, client lib).

• aztec-cli (currently empty)
• aztec.js
• aztec-node
• aztec-rpc

Modules

modules folder, contains packages each with a responsibility specific to an aztec3 module (sequencer, simulator, etc), which power the apps.

• acir-simulator
• archiver
• kernel-prover (single file) -> Consider moving into aztec-rpc
• key-store (currently empty)
• p2p
• prover-client (currently empty)
• sequencer-client
• world-state

Wrappers

wrappers folder, packages that wrap imported functionality.

• bb.js
• circuits.js
• l1-contracts

Libraries

libs folder, contains generic shared code or base models.

• ethereum.js
• merkle-trees
• foundation
• noir-contracts -> Unsure what exactly goes in here, maybe it's a module?
• types

Other:

misc folder, everything else.

• docs
• end-to-end
• yarn-project-base -> Could be kept at the root of the tree?

[spalladino]

Given lack of traction, and how disruptive a change like this is, I'm tempting to close this one as wontfix. Thoughts @ludamad @PhilWindle @rahul-kothari..?

[rahul-kothari]

Tempted to agree - bigger fish to fry for now it seems

[PhilWindle]

I agree, leave for now.