Add DisplayAsDebug - A struct to help with Debug impls

[Centril]

This adds a simple wrapper:

pub struct DisplayAsDebug<T>(pub T);

impl<T: fmt::Display> fmt::Debug for DisplayAsDebug<T> {
    fn fmt(&self, fmt: &mut fmt::Formatter) -> fmt::Result {
        <T as fmt::Display>::fmt(&self.0, fmt)
    }
}

r? @alexcrichton

[sfackler]

We can be a bit more flexible here and be generic over T: Display.

[Centril]

@sfackler Interesting...! so rename to DebugDisplay?

[Centril]

And now we are more flexible with DebugDisplay ;)

[sfackler]

I've written this a couple of places when I've needed to do something like this. It's a tiny little utility, but I think nice to have.

@rfcbot fcp merge

[rfcbot]

Team member @sfackler has proposed to merge this. The next step is review by the rest of the tagged teams:

• @Kimundi
• @SimonSapin
• @alexcrichton
• @aturon
• @dtolnay
• @sfackler
• @withoutboats

Concerns:

• exists (Add DisplayAsDebug - A struct to help with Debug impls #49067 (comment))

Once a majority of reviewers approve (and none object), this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up!

See this document for info about what commands tagged team members can give me.

[SimonSapin]

Bikeshed: DebugAsDisplay?

[sfackler]

Yeah, I was kind of thinking that an As in the middle would be good as well.

[Centril]

Totally agree with the As bit;

Per discussion on #rust-libs I went with DisplayAsDebug since you are "converting" a Display impl in effect to a Debug impl (at least I think that reads somewhat more naturally than the reverse).

I took the liberty of making a tracking issue now so I don't have to make a commit again ^,-

[kennytm]

Why can't you use format_args! for this

use std::fmt::{Debug, Formatter, Result};

struct Arm<'a, L: 'a, R: 'a>(&'a (L, R));
struct Table<'a, K: 'a, V: 'a>(&'a [(K, V)], V);

impl<'a, L: 'a + Debug, R: 'a + Debug> Debug for Arm<'a, L, R> {
    fn fmt(&self, fmt: &mut Formatter) -> Result {
        L::fmt(&(self.0).0, fmt)?;
        fmt.write_str(" => ")?;
        R::fmt(&(self.0).1, fmt)
    }
}

impl<'a, K: 'a + Debug, V: 'a + Debug> Debug for Table<'a, K, V> {
    fn fmt(&self, fmt: &mut Formatter) -> Result {
        fmt.debug_set()
           .entries(self.0.iter().map(Arm))
           .entry(&Arm(&(format_args!("{}", "_"), &self.1)))
           .finish()
    }
}

fn main() {
    let table = (1..3).enumerate().collect::<Vec<_>>();
    assert_eq!(format!("{:?}", Table(&*table, 0)),
               "{0 => 1, 1 => 2, _ => 0}");
}

[alexcrichton]

@rfcbot concern exists

Thanks @kennytm! Looks like this may already exist in libstd?

[Centril]

Some thoughts:

1. I had no idea that Arguments had this behavior -- so this was entirely unexpected. Is this a guarantee of format_args! / Arguments to behave in this way tho (or just a happy accident)? If so - it should be documented somewhere.

2. The specific case above can be shortened to: format_args!("_").

3. There's a minor overhead: 16 bytes vs. 48 -- it's not much (so it probably does not warrant an extra type), but I thought I'd mention it.

[SimonSapin]

Interesting! I think I’m now in favor of closing this RFC and either adding a new example or modifying existing examples to show this use of format_args! in the docs of fmt::Formatter::debug_{struct,tuple,list,set,map} and fmt::Debug{Struct,Tuple,List,Set,Map}.

[dtolnay]

👍 for directing people to format_args! for this use case. @rfcbot fcp close

[rfcbot]

@dtolnay proposal cancelled.

[rfcbot]

Team member @dtolnay has proposed to close this. The next step is review by the rest of the tagged teams:

• @Kimundi
• @SimonSapin
• @alexcrichton
• @aturon
• @dtolnay
• @sfackler
• @withoutboats

No concerns currently listed.

Once a majority of reviewers approve (and none object), this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up!

See this document for info about what commands tagged team members can give me.

[Centril]

I'll expedite the process and close the PR myself ;)