Tom Fryers

[Here’s](https://github.com/psf/black/runs/5278179500?check_suite_focus=true#step:11:1) diff-shades’ report when the new style is set as the default (rather than hiding it behind `--preview`). It’s +15 274 / ‒18 914 out of 2 136 881 lines.

The most recent commit addresses an example of the single-line docstring line-length issue. It’s a bit complicated. Re-reading PEP 8, it seems the 72-character limit does not apply to docstring...

> * The one specific choice in this PR that I'm having trouble with is putting the opening quotes on their own line. It's clear that in Black itself previously...

[diff-shades with the same-line opening quotes style](https://github.com/psf/black/runs/5388767567?check_suite_focus=true#step:11:1) is +13 296 / ‒29 637 out of 2 147 073. Testing this led me to discover an issue with this quote-placement style:...

The tests seem to have stopped working. I don’t really have time to fix them right now, but probably will in a month or so.

I think what is meant is that Black should only break lines inside brackets if it results in sufficiently shorter lines. For example, Black currently formats ```python f(9876543210987654321098765432109876543210987654321098765432109876543210987654321098765432109876543210) ``` to...

I had a look for cases of this in some of my code, and I think long string literals are by far the most common cause, and they’ll cease to...

Or, even better, `my_dict = my_dict.setdefault(key, {})`.

> frozen += frozen # raises AttributeError > frozen += mutable # raises AttributeError Why not have these be equivalent to `frozen = frozen + frozen` and `frozen = frozen...

I see your point re preventing accidental reassignments. My worry is that this is different behaviour to all of the built-in immutable types, which may be unexpected.