Tall style: Regression for block-style LHS

[nex3]

The tall style formats the following code as so:

void main() {
  for (var i = 0; i < _importers.length; i++) {
    for (var j = 0; j < i; j++) {
      _perImporterCanonicalizeCache[(
            _importers[j],
            url,
            forImport: forImport,
          )] =
          null;
    }
  }
}

where previously it was formatted as

void main() {
  for (var i = 0; i < _importers.length; i++) {
    for (var j = 0; j < i; j++) {
      _perImporterCanonicalizeCache[(
        _importers[j],
        url,
        forImport: forImport,
      )] = null;
    }
  }
}

The former seems worse to me, and the latter doesn't seem obviously contrary to the tall style.

[munificent]

Oh, wow, interesting. Yeah, the new style doesn't recognize an index operator expression as a being block-formattable, which is why you get the extra indentation. I had never even considered that you might have a large enough index operand to cause it to split on the LHS.

For what it's worth, if this were my code, I'd probably hoist the record literal out to a local variable because I find this fairly hard to read in either formatting style. But I agree the tall style output isn't ideal.

[nex3]

I've always been a bit more LISP-brained than you 🙂