filter empty statements when checking block emptiness

[calebcartwright]

Fixes #3868

[topecongiro]

@calebcartwright Why do we need to update all the tests?

[calebcartwright]

Sorry i meant to leave a note on that. Apparently both ; and (); are parsed as empty tuple exprs in the block statements, so with the updated logic filtering out empty statements, these blocks were now being detected as empty.

    loop {
        ();
        ();
        }

would be flattened to

    loop {}

which was making these tests fail since they're trying to validate brace placement. the updates were just to convert to non-empty statements in those blocks to prevent the flattening.

Is this the correct behavior? Should we treat a block that only has a (); statement the same as one that only has a ; (both as empty blocks)?

If not I can do some minor refactoring to get the SnippetProvider available when checking the block statements, and in the event of the empty tuple grab the associated snippet and check for ; vs ();

[topecongiro]

Thank you for your explanation!

Is this the correct behavior? Should we treat a block that only has a (); statement the same as one that only has a ; (both as empty blocks)?

Removing (); seems fine to me, though I am afraid that we may need to version-gate this as we didn't use to remove it. As we will be preparing for 2.0 release in the near future, I would like to merge this PR after we released 1.4.10 so that we don't need to add an extra version-gate which will be removed soon.

[calebcartwright]

SGTM!

[topecongiro]

@calebcartwright Sorry, would you mind resolving the conflict? I should have merged this before #3880.

[calebcartwright]

@calebcartwright Sorry, would you mind resolving the conflict? I should have merged this before #3880.

Sure, will do!

[topecongiro]

Thank you for the update!