Refactor printing inline type declarations

[zth]

Fixes #7737

EDIT: The below has been fixed.

@shulhi separate issue, but I've noticed for inline type declarations, the formatted automatically "compacts" the records. I wonder if this is something we could fix somehow? It's above my syntax knowledge.

Examples:
1) In this PR you'll see that type.resi defines the nested record as:

type options = {
  permissions: {
    all: bool
  }
}

But it's printed as:

type options = {
  permissions: {all: bool},
}

Similarily, if I change it one more level and define the type in type.resi as:

type options = {
  permissions: {all: bool}
}

Then it again compacts it one level:

type options = {permissions: {all: bool}}

I wonder if this is some easy fix or some setting I've just missed in printing. Anyone have ideas?

[zth]

The formatting issue lies in how force_break is calculated. It's looking at lines of labels from parsing, but these lines won't be aligned properly with inline records. Need to figure out how to deal with that.

[shulhi]

The unstable formatting is related to #6976

[nojaf]

Could you update the snapshots so we know what has changed by this fix?

[zth]

Could you update the snapshots so we know what has changed by this fix?

I've added to an existing test file, which you can see in the diff. Separately, there's an issue described in the PR about formatting not working as intended, which is what's making CI red. So that needs either a proper or intermediate fix. I have some ideas.

[pkg-pr-new]

Open in StackBlitz

rescript

npm i https://pkg.pr.new/rescript-lang/rescript@7741

@rescript/darwin-arm64

npm i https://pkg.pr.new/rescript-lang/rescript/@rescript/darwin-arm64@7741

@rescript/darwin-x64

npm i https://pkg.pr.new/rescript-lang/rescript/@rescript/darwin-x64@7741

@rescript/linux-arm64

npm i https://pkg.pr.new/rescript-lang/rescript/@rescript/linux-arm64@7741

@rescript/linux-x64

npm i https://pkg.pr.new/rescript-lang/rescript/@rescript/linux-x64@7741

@rescript/win32-x64

npm i https://pkg.pr.new/rescript-lang/rescript/@rescript/win32-x64@7741

commit: 28cc10c

[zth]

Ok @shulhi I think I cracked it for this specific case (printing inline records). The problem was that we were only caring about the label locs when deciding whether to break in formatting or not. That's fine for regular records, but when printing inline we want to care about the full inline record body to decide whether to break or not, since that is what has the information about how the inline records were laid out in the code at parse time.

[zth]

Move the inline record printing logic into print_type_declarations itself. That's where it belongs IMO.

[zth]

Special case printing inline records - look at the full inline record loc to decide whether the code had line breaks (and should get force breaks) or not.

[nojaf]

Thanks!