Expand `corepack use` documentation description

[MikeMcC399]

Issue

Under README > corepack use the complete description currently says:

corepack use <name[@]>

When run, this command will retrieve the latest release matching the provided descriptor, assign it to the project's package.json file, and automatically perform an install.

package.json location

There is no description of how the command navigates the directory hierarchy to locate a possible existing package.json to modify.

Using the test example of @arcanis in #607 (comment) shows that it operates on the root of a hierarchy:

corepack enable yarn
cd $(mktemp -d)
yarn init -2 -w
mkdir -p packages/foo
echo '{}' > packages/foo/package.json
jq .packageManager package.json
(cd packages/foo && corepack use yarn@4.5.0)
jq .packageManager package.json

In this example case, the result is that:

• packages/foo/package.json contains {}

• package.json contains

  {
    "name": "tmp.IcgYOXK7aS",
    "packageManager": "yarn@4.5.0+sha512.837566d24eec14ec0f5f1411adb544e892b3454255e61fdef8fd05f3429480102806bac7446bc9daff3896b01ae4b62d00096c7e989f1596f2af10b927532f39",
    "workspaces": [
      "packages/*"
    ]
  }
  

package.json creation

If no package.json file is found, then one is created. This effect of the command is not described.

Suggestion

For the README > corepack use section:

• Add text to explain how the location for the package.json file is determined.
• Add text which describes how a package.json file and / or a packageManager field is created if they don't exist.

Related

• corepack use in monorepo targets root not project package.json #607