[Documentation] Additional documentation needed in the Yarn 2.0 migration related to "script" incompatibilities

[yinzara]

• I'd be willing to implement this feature
• This feature can already be implemented through a plugin

Describe the user story

With the introduction of the yarn shell in Yarn 2.0, there are actually many idiosyncrasies about differences between how a script may function in Yarn 1 and Yarn 2. From glob pattern auto-expansion to non-existent environment variable references causing errors, there are a set of things a Yarn 1.0 user needs to know to ensure his "script"s remain working in Yarn 2.0

The Yarn 2.0 migration page at https://yarnpkg.com/advanced/migration mentions calling your scripts via "yarn node" and the details about pre and post scripts, but makes no mention of the incompatibility of the scripts themselves. I did not discover these issues until I was already fully converted with all other changes and spent more time fixing my scripts than I did with almost everything else in the migration combined.

Describe the solution you'd like

A section in the migration page about the general things to look out for in your scripts and ways to correct them. Simply knowing I could replace "$PWD" with "$(cwd)" or escaping anything glob pattern like that didn't match files with single quotes, would have been extremely helpful and saved me posting issues.

[arcanis]

Yup, good point 👍

[jeffrson]

Yes, this would be very helpful!

Use case: we have a package.json script, that calls node -e "..." with a regular expression including "^".

Apparently yarn breaks the input line at that position. How can we solve the error (without an additional script)?