Tall style formatter hangs on certain string interpolations

[Mike278]

I tried running the new formatter on my project and it got stuck on a certain file. I eventually narrowed it down to some specific string interpolation code. For example running dart format --enable-experiment=tall-style example.dart with SDK 3.5.0

var _ = '''
${x(1).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
${x(1).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
${x(1).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
${x(1).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
${x(1).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
${x(1).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
${x(1).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
${x(1).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
''';

You can play with the number of rows and columns, but the important bits seem to be:

• must contain at least 2 function calls, each with at least 1 argument
• must overflow the 80 character limit

Playing with the above input:

• 4 lines: 300ms
• 5 lines: 1180ms
• 6 lines: 6200ms
• 7 lines: 44360ms
• 8 lines: left this one blank in case it completed by the time I was done writing this post... it did not 🤪

Note: the formatted output for 4 lines looks like:

var _ = '''
${x(
  1,
).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
${x(
  1,
).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
${x(
  1,
).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
${x(
  1,
).x(1).xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx}
''';

FWIW, the original code is a generated SQL statement that glues together different query fragments via function calls.

[munificent]

Thanks! Yeah, I think large strings with complex interpolation expressions are still a corner where the solver can go pathological. I need to think more about how to handle that.