Refactor to use debug_struct in several Debug impls

[MaloJaffre]

Also use pad and derive Debug for Edge.

Fixes#44771.

[rust-highfive]

r? @sfackler

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

[ollie27]

The {{ and }} need to be changed to { and } as this string is no longer being used as a format string.

[MaloJaffre]

Good catch, thanks for the review!
I removed the double braces.

[sfackler]

I don't think this really fixes what #44771 was talking about - the output is still going to be on a single line if you format with {:#?}.

[MaloJaffre]

I've tested on my machine that:

println!("{:#?}", std::sync::Mutex::new(5));

prints (because of debug_struct use in Debug impl):

Mutex{data:5}

But you're right that for e.g. mpsc::Sender, the output is only on one line, because I didn't use debug_struct (I must only output .., without a field name).

I think this is another related issue that we can fix later, because I just refactored this part to be more consistent with other Debug impls.

[carols10cents]

ping @sfackler - do you think this can be merged as-is to fix part of the problem?

[sfackler]

The proper change seems like fmt.debug_struct("Sender").finish() rather than fmt.pad("Sender { .. }").

[MaloJaffre]

Using fmt.debug_struct("Sender").finish() will show Sender instead of Sender { .. }, is that an acceptable change of output?

[sfackler]

Yep, debug representations aren't stable.

[MaloJaffre]

I've rebased and applied the requested changes.

Mutex now uses a LockedPlaceholder struct with a custom Debug impl, like RefCell.

With these changes,

fnmain(){let a = std::sync::Mutex::new(5);let b = a.lock().unwrap();println!("{:#?}", a);}

prints:

Mutex{
data: <locked>
}

[sfackler]

Thanks!

@bors r+ rollup

[bors]

📌 Commit 7b02a2d has been approved by sfackler

[kennytm]

@bors r-

CI failed.

[01:14:12] thread '<unnamed>' panicked at 'assertion failed: `(left == right)`
[01:14:12] left: `"Handle"`,
[01:14:12] right: `"Handle { .. }"`', /checkout/src/libstd/sync/mpsc/select.rs:789:8
[01:14:12] thread '<unnamed>' panicked at 'assertion failed: `(left == right)`
[01:14:12] left: `"Select"`,
[01:14:12] right: `"Select { .. }"`', /checkout/src/libstd/sync/mpsc/select.rs:781:8
[01:14:13] thread '<unnamed>' panicked at 'assertion failed: `(left == right)`
[01:14:13] left: `"Receiver"`,
[01:14:13] right: `"Receiver { .. }"`', /checkout/src/libstd/sync/mpsc/mod.rs:3022:8
[01:14:13] thread '<unnamed>' panicked at 'assertion failed: `(left == right)`
[01:14:13] left: `"Sender"`,
[01:14:13] right: `"Sender { .. }"`', /checkout/src/libstd/sync/mpsc/mod.rs:3016:8
[01:14:13] thread '<unnamed>' panicked at 'assertion failed: `(left == right)`
[01:14:13] left: `"SyncSender"`,
[01:14:13] right: `"SyncSender { .. }"`', /checkout/src/libstd/sync/mpsc/mod.rs:3028:8
...
[01:14:23] failures:
[01:14:23] sync::mpsc::select::tests::fmt_debug_handle
[01:14:23] sync::mpsc::select::tests::fmt_debug_select
[01:14:23] sync::mpsc::sync_tests::fmt_debug_recv
[01:14:23] sync::mpsc::sync_tests::fmt_debug_sender
[01:14:23] sync::mpsc::sync_tests::fmt_debug_sync_sender
[01:14:23] [01:14:23] test result: FAILED. 803 passed; 5 failed; 0 ignored; 0 measured; 0 filtered out

[MaloJaffre]

Sorry for the failed CI, this should be fixed now.

[sfackler]

Can we just get rid of those tests? It doesn't seem like they're doing anything useful.

[MaloJaffre]

Sure, I've removed these tests and squashed commits.

[sfackler]

@bors r+ rollup

[bors]

📌 Commit 679457a has been approved by sfackler