core: fmt: debug builders pretty print tuple structs and enum variants in a single line

[liigo]

Before this PR, println!("{:#?}", Some(1)); pretty prints debug output as:

Some(
1
)

(by adding newlines and indent)

But I don't think it's pretty print: it's weird, even not a preferred code style.

Some(1) itself (without newlines and indent) is pretty print, IMO.

This PR change the pretty print output, to print tuple structs and enum variants in a single line. (We already pretty print tuples in this way.)

Perhaps this is a [breaking-change]? And an RFC is required? I don't known.

[rust-highfive]

r? @aturon

(rust_highfive has picked a reviewer for you, use r? to override)

[alexcrichton]

r? @sfackler

[sfackler]

The current behavior is intentional, as described in the RFC: rust-lang/rfcs#640. The pretty printing mode is designed for ease of readability, not matching rust style. Printing some things in one line and others in multiples does not nest well:

MyTupleStruct(MyStruct{a:1,b:TupleStruct(MyOtherStruct{foo: bar
},10,"hi",Thing{baz: buzz
})},true,false,OtherStruct{thing:"hi"})

It could potentially be nice to print single-variant tuples in one line if there are no line breaks in the output for the inner value, but I'm not sure how that would work.

[liigo]

@sfackler

The current behavior is intentional, as described in the RFC: rust-lang/rfcs#640

OK, maybe an RFC is required to propose changing this a bit?

The pretty printing mode is designed for ease of readability, not matching rust style.

But I do think that Some(1) is more readable than

Some(
1
)

Printing some things in one line and others in multiples does not nest well

Tuples with nested struct, already pretty print in a single line, and this PR doesn't change that. http://is.gd/ZJKuVy

[sfackler]

The tuple behavior is a bug, not a feature.

[liigo]

So, let's resolve it through an RFC

[liigo]

I've written an RFC rust-lang/rfcs#1198

[bors]

☔ The latest upstream changes (presumably #26913) made this pull request unmergeable. Please resolve the merge conflicts.

[alexcrichton]

The associated RFC was closed, so I'm going to close this as well.