feat(core): support for 3rd party package managers

[Jordan-Hall]

Current Behavior

Currently you are limited to only the packages provided by NX repo and it does not allow for extending without NX support.

Major change is installPackagesTask now returns a promise instead of a void.

Expected Behavior

You can now create a plugin that manages package. Such as deno, bun and china package managers

Related Issue(s)

Fixes #15622

[nx-cloud]

☁️ Nx Cloud Report

CI is running/has finished running commands for commit 42ec3b9. As they complete they will appear below. Click to see the status, the terminal output, and the build insights.

📂 See all runs for this branch

---

⌛ The following target is in progress

• nx affected --target=e2e --base=74c812e2e88a1770e2fe4923c24c3bdd7489e508 --head=42ec3b95045df7c7f01aa18c68ca45e15b097d95 --parallel=1

✅ Successfully ran 7 targets

• nx affected --target=test --base=74c812e2e88a1770e2fe4923c24c3bdd7489e508 --head=42ec3b95045df7c7f01aa18c68ca45e15b097d95 --parallel=1
• nx affected --target=build --base=74c812e2e88a1770e2fe4923c24c3bdd7489e508 --head=42ec3b95045df7c7f01aa18c68ca45e15b097d95 --parallel=3
• nx run-many -t check-imports check-commit check-lock-files check-codeowners documentation --parallel=1 --no-dte
• nx affected --target=lint --base=74c812e2e88a1770e2fe4923c24c3bdd7489e508 --head=42ec3b95045df7c7f01aa18c68ca45e15b097d95 --parallel=3
• nx-cloud record -- nx format:check --base=74c812e2e88a1770e2fe4923c24c3bdd7489e508 --head=42ec3b95045df7c7f01aa18c68ca45e15b097d95
• nx documentation --no-dte
• nx affected -t e2e-macos --parallel=1 --base=74c812e2e88a1770e2fe4923c24c3bdd7489e508 --head=42ec3b95045df7c7f01aa18c68ca45e15b097d95

---

Sent with 💌 from NxCloud.

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
nx-dev	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Aug 29, 2023 9:00pm

[michaelangeloio]

+1 on this

[ffMathy]

I can't believe no one has looked at this yet 😒

@Jordan-Hall do you think it'll help them merge it if you solved the conflicts?

I hope we'll even hear from the team.

[Jordan-Hall]

@ffMathy there was some discussion on the community slack/discord about this. @AgentEnder has some ideas but it's not high on priority for the team. Don't think it was just ignore the team really does engaged with community

[ffMathy]

Could those ideas be shared so we can contribute, and they only have to press the merge button? 😅

[Jordan-Hall]

Craigory:

Yeah, its not something that we've thoroughly discussed. Its not something that super makes sense given that there's not a ton of PM's out and about, and very very few with even 1% of the market share
Additionally, things that are set on this incorrectly could very easily cause errors on commands like nx migrate or when running generators that install packages, which would then get reported as issues and so on
Further, there's a lot to do with package managers in regards to how we parse lock files to add nodes and deps to the graph... so if these new package managers add a new lockfile type then that needs to be handled as well

Miro:

We plan to move the entire package manager (including lock file logic) to @nx/js package and have it entirely optional, so that would open the door to any new package managers.

But as craigory mentioned, the new package manager's logic would have to include several tricky implementations - namely lock file parsing and pruning. Also, given the reach these package managers now have making changes to Nx core to support this would have low priority.

It's an interesting requirement, though. The PR is going in the right direction, but there's way more ground that needs to be covered to support something like this.

I would consider helping with move to js package. That's a good idea tbh. But that requires alot of planning and architecture design which team likely not decided

[ffMathy]

But, for now, given that Bun's implementation is identical to Yarn, couldn't we just make an easy fix temporarily while we wait for the rest?

I get the impression that Bun's package manager won't be less than 1% market share for very long.

[jaysoo]

Closing this PR as it requires additional design discussion. Added this new discussion to keep track of the conversation.

[github-actions]

This pull request has already been merged/closed. If you experience issues related to these changes, please open a new issue referencing this pull request.