Add match_arm_indent option

[kaathewisegit]

Allows to disable the indentation of match arms.

Related issue: #2937

[ytmimi]

Thanks for taking the time to work on this older feature request.

The code seems pretty straightforward to me, but I've left some comments inline. Take a look when you've got a chance.

[kaathewisegit]

Applied the changes, moved the tests to the configs/match_arm_indent folder, and added a bunch more tests. I also remembered that this feature interacts with nested matches, so I added a few cases for that

[ytmimi]

Applied the changes, moved the tests to the configs/match_arm_indent folder, and added a bunch more tests. I also remembered that this feature interacts with nested matches, so I added a few cases for that

Thanks for making the changes and expanding on the test case. I'll have some time to re-review this early next week.

[ytmimi]

Good test case, but it's pretty unfortunate formatting. It's really hard to tell that these are nested match statements.

[kaathewisegit]

Yeah, that's the main problem with this formatting option. The example is deliberately obtuse, but I think it's somewhat better with realistic usage. That's how I envision it working for AST matching, for example:

match statement {
Stmt::Local(item) => match item {
Item::Const(item_const) => {
    // do stuff with const
},
Item::Enum(item_enum) => {
    // do stuff with enum
},
// rets of items
},
Stmt::Expr(expr, semi) => match expr {
Expr::Array(arr) => {
     // do stuff with arr (and semi, with is in scope despite not being in the second match statement)
},
Expr::Assign(assign) => {
    //  ...
},
// ...
},
}

Here the flat structure is desirable, as we're processing all of the items together.

In theory, there could be a heuristic which would wrap the nested match as a body, indenting it. But I'm not sure how it could be applied so that it both fixes these edge cases, but doesn't break sane nesting

[ytmimi]

I went ahead and created a tracking issue for this #6533. When you get a chance please rebase the PR and add the tracking issue link to the option's documentation.

[kaathewisegit]

Added the link to the tracking issue

[ytmimi]

Overall the implementation is pretty straightforward and doesn't complicate match arm formatting. I still think the nested cases are hard to read, but I'd say we're fine to move forward with this as an unstable option.

[kaathewisegit]

Thanks for taking the time to review it!

[ytmimi]

Just an FYI the new option won't be available on nightly until the next subtree sync with rust-lang/rust. hoping to get that out soonish

[kaathewisegit]

Oh, yeah, I'm not in a hurry. I know that things take time when it comes to releases. I'm actually on stable myself, so I was hoping that someone on nightly would pick up this feature and test it in the meanwhile