Something I wonder though: can we have automated tests that make some smoke test with prettier v2 and v3, to check that everything works?

cc @sosukesuzuki @fisker do you know if some plugin is doing something like that?

I don't understand all the changes here, but since we've already released v1.0.0, I'm fine with merging this!

So mainly V2 would still generate the concat object under the hood so in some places when we print comments, with V2 we would get a concatenation of comments and with V3 we would get an array of comments. We needed an array always so we could iterate, that's why we have to check if it's V2 so we can retrieve the array within the concat object, while in V3 we just ask for the array.

other changes is that the clean function is more strict now and doesn't iterate over comments. so now we have to clean the comments as well.

and for testing, prettier now returns promises so we have to wait for the resolution.

I also added a utility function that tells us if the prettier version satisfies a range. that way we can use the same function to check if we are in V2 or V3 and also if we are at least V2.3.0 which is when we had to force our users to update their prettier version.

also since there are branches that we would never reach because one thing if mocking Prettier version on tests and another is Mocking the behaviour of the concat builder (or lack of it) so I had to lower our coverage standards.

prettier standalone plugins file names also changed in V3 so we had to adapt so our latest markdown would be properly loaded on standalone testing.

and finally the thing that started this rabbit hole. The parse function now receives 2 arguments and in V2 we used to get 3. So we had to adapt.

Something I wonder though: can we have automated tests that make some smoke test with prettier v2 and v3, to check that everything works?

in this PR there's a new task in our CI that installs V2 before running the tests in V2. I'm not too familiar with GitHub actions but I assume we could run all of the tests only if a particular small test fails.

in this PR there's a new task in our CI that installs V2 before running the tests in V2. I'm not too familiar with GitHub actions but I assume we could run all of the tests only if a particular small test fails

I missed that part, that is cool and I think it's good enough!

And thanks for the explanation 😃