inline comments make subscription look really bad

[wimglenn]

input:

my_value = my_list[0]  # developer wanted to explain something here with an inline comment

actual output (v19.10b0):

my_value = my_list[
    0
]  # developer wanted to explain something here with an inline comment

desired output:

# developer wanted to explain something here with an inline comment
my_value = my_list[0]

Instead of wrapping long lines so weird, do you think it would be feasible to just move these comment above instead?

[zsol]

It's hard to do in the general sense because the comment might have some meaning attached to that line (think code analysis tools like flake8 or mypy), and so moving the comment might change that meaning.

[ocehugo]

Related to #1187

I don't know the internals, but if other code analysis tools got simple functions to return all possible pragmas, a feasible solution is to add those to a "dont split inline comment" rule.

[sebastian-fredriksson-bernholtz]

@zsol The existing behaviour creates the same issue in many cases - if not most/all.

I really think black should leave inline comments be. As I described in #1187 (comment) inline comments are probably inline for a reason. Code analysis tool is a very strong argument.

[th0ger]

Ok, inline comments should stay inline. But are inline comments handled consistently?

# in:

a = 1/3  # Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor
b = (
    0.333333333333333333333333333333333333333333333333333333333333
) # Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor

# out:

a = (
    1 / 3
)  # Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor
b = 0.333333333333333333333333333333333333333333333333333333333333  # Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor

Here, Black is clearly doing a terrible job in reducing the max line length. The introduction of parenthesis for a could have been applied to (or kept for) b as well, reducing line length significantly.