Upgrading Yarn to V2 and Dropping Lerna

[spacesailor24]

• 1.x
  uses
  - npm
  - lerna (in fixed mode)
• 4.x
  uses
  - yarn: package manager, with workspace support (handles local package linking and not duplicating package dependency installs)
  - lerna (in independent mode): handles running package.json scripts for each package (can be replaced by yarn workspaces), but does not handle package versioning and publishing (currently it's a manual process)

---

These are the decisions that need to be made for 4.x:

1. Are we upgrading to Yarn v2?
   • Judging solely from this article, it would seem wiser to stick with v1
2. Are we dropping the use of Lerna?
   • Regarding running package.json scripts for each package, Yarn workspaces provides this feature
   • If we decide to revert the decision of using the independent method (and use fixed as in 1.x), we get the benefit of Lerna managing updating package versions and publishing each package to NPM
     • Keep in mind that the decision to move away from this was because a single breaking change in one package means all version numbers for every package must have a major version bump (even if they aren't affected by the breaking change)
   • If we stay with independent, it is my naive recommendation to drop the use of Lerna and just write simple scripts to update package versions and publish to NPM (although I have no idea what kind of can of worms may be awaiting us)
     • Talking with @mpetrunic, Lodestar used the independent method and they found it to be a headache to deal with updating package versions for each dependent package, and decided to revert to fixed

[mpetrunic]

Judging solely from this article, it would seem wiser to stick with v1

That pretty outdated article, it was testing first release candidate of yarn2 :)
https://spin.atomicobject.com/2021/02/17/switch-yarn-2/
https://www.sitepoint.com/yarn-vs-npm/