Modernize our usage of Lerna (our monorepo solution)

[Skaiir]

What should we do?

There have been tons of changes to lerna following it being picked up, and while we've considered migrating away from it in the past, it seems that following it's change of maintainer, the library is thriving again. So we should at the very least update to introduce some features which would bring us some value.

Two features that would bring us some good value would be:

• caching of builds / test results when nothing about a subrepo changes
• using the lerna provided proper build pipeline configuration, instead of handling all the logic in package.json scripts

Additionally we should fix the fact that every library gets built unnecessarily whenever we run any project.

Why should we do it?

This all would allow us to overall build less, and build more parallelized, which may be a significant improvements in terms of developer experience and build times across the board.

[nikku]

👍 Core modeling uses latest version of Lerna + npm workspaces and it works fine for us.

[Skaiir]

@nikku Do you also use the lerna pipelines, caching and all those good features that have been added recently? It's been super useful to reduce the amount of redundant rebuilds and retests.

[nikku]

We don't, and we don't see the need (yet), too. But happy to hear your feedback what works and what does not.