I'd expected that adding the --experimental-cli flag would generate the same format as without it. So I was surprised that when I ran it on a monrepo with about 1000 files, I got about 100 files with changes.

pnpm exec prettier --write ./scripts 'packages/**'

[email protected] and [email protected] give the same output.

with:

pnpm exec prettier --write ./scripts 'packages/**' --experimental-cli  

[email protected] gives lots of changes.

with things like:

const UserAvatar = (props: {
  username: string;
  displayProfilePic: boolean;
}) => {
  const theme = useTheme();

becoming:

  const UserAvatar = (props: { username: string; displayProfilePic: boolean }) => {
  const theme = useTheme();

and:

  <Link
    to={example.link || `/${example.route}`}
    component={CardActionArea}
    target="_blank"
  >

becoming:

  <Link to={example.link || `/${example.route}`} component={CardActionArea} target="_blank">

I have a very simple config (.prettierrc.cjs):

module.exports = {
  trailingComma: 'es5',
  tabWidth: 2,
  endOfLine: 'auto',
  singleQuote: true,
  printWidth: 120,
};

I'm not saying that I particularly mind the changes, though I was used to the previous result, but I didn't expect that changing the flag would have such a side effect.

As for performance, it's massively faster. Previously I ran prettier in each sub package because it was faster that way. That is no longer true.

So the experimental flag is a huge perf improvement and simplifies my setup nicely.