community-plugins/docs/updating-yarn.md

Updating Yarn

Overview

To keep workflows and tooling easy to maintain, we use a unique yarn version across all workspaces. This document outlines the process for updating the yarn version across all workspaces.

Process

[!NOTE] this is the process used from version 3.x to 4.x, future version could imply a different process.

1. Update Yarn Version

update the yarn version runing this command on the root of the repository:

yarn set version latest # or a specific version if is needed

2. Update Workspaces

Bumping from version 3 to 4 implies changes on all the workspaces yarn.lock files, to update all the following command can be used:

chmod +x ./scripts/run-in-all.sh
./scripts/run-in-all.sh "yarn install && yarn dedupe --mode=update-lockfile && rm -rf node_modules"

What this command does is to iterate over all the workspaces and run yarn install on each one, full install is needed for some versions when yarn changes the checksums (i.e. from v3 to v4). The yarn dedupe command is used to update the yarn.lock file, this command will update the yarn.lock file with the new checksums and dependencies. The rm -rf node_modules command is used to remove the node_modules folder, this is optional but have in mind installing node modules in all golder will consume more than 100gb on disk.

If for some version update the --mode=update-lockfile flag can be used to update the yarn.lock file only saving time form yarn link and other steps.

3. Create a PR

After running the previous command, a PR should be created with the changes on the yarn.lock files. This PR will affect all the workspaces but maintainer will only wait for a couple of approvals to merge it.