Determine if ASTGen is responsible for checking that optional tokens exist in the SwiftSyntax tree

[ahoppen]

https://github.com/apple/swift/pull/67111/files#diff-de110680caa5f6af695d85ee44bc75ce4693b169e6e537ca176ec3fda17705a8R22-R29 checks that a FunctionCallExprSyntax has a left parenthesis and a right parenthesis if there are arguments. We need to figure out if it’s ASTGen’s responsibility to performs such checks.

Options are:

1. We change the SwiftSyntax tree to make sure that the parenthesis are always present if one of the following conditions is satisfied (which might be tricky)
   a. There is a single argument in arguments or
   b. There is no trailing closure
2. ASTGen performs these checks like it does today
3. Since the SwiftSyntax tree is structurally valid without the left and right parenthesis, ASTGen does not need to check these invariants unless it requires the parentheses e.g. to get their locations.

I’m favoring option (3) at the moment

[saehejkang]

If ASTGen does not need to make the check for the other invariants, is this issue just a simple deletion of that check? The left paren and right paren are passed to the visit, so only a check there would be appropriate?

[bnbarham]

is this issue just a simple deletion of that check

The short answer here is really that we haven't decided yet.

The currently-marked ASTGen issues are mostly from tasks that we identified in #67111 as needing to happen before any further work in this area. In some cases (like this one), we haven't necessarily decided on a direction yet. Other than the issues marked good first issue, they're generally fairly involved and were mostly opened so that we didn't lose the discussion in that PR.

@rintaro and @hamishknight will be working in this area, so I'll go through and assign them appropriately. If you're interested in following the work, we'll make sure to mark any related PRs with ASTGen :)

[saehejkang]

are there any other good first issues or other issues in general that will be good to work on for gaining a better understaniding Looking at the good first issues that are open, a lot of them are really old or being worked on.

[bnbarham]

Sorry @saehejkang, missed this comment. I don't think there's any in ASTGen at the moment (pending some foundational work), but it's likely we'll have more in the future and we'll file/mark them then :)