Tracking issue for hexadecimal integers with fmt::Debug output

[alexcrichton]

This is a tracking issue for rust-lang/rfcs#2226, @SimonSapin mind filling in some details here?

[SimonSapin]

Implementation PR: #48978
RFC: https://rust-lang.github.io/rfcs/2226-fmt-debug-hex.html

In formatting strings (in format! and related macros), this adds two "formatting types" x? and X? (in addition to the existing x, ?, p, etc) that make integers be formatted in (lower- or upper-case) hexadecimal with the fmt::Debug trait. This is useful when a composite type implements fmt::Debug (for example a struct with #[derive(Debug)]) but not fmt::LowerHex or fmt::UpperHex, and contains integers or a slice of integers or a byte string.

For example:

assert!(format!("{:02x?}", b"Foo\0") == "[46, 6f, 6f, 00]");
assert!(format!("{:02X?}", b"Foo\0") == "[46, 6F, 6F, 00]");

[shepmaster]

I assume this should be closed now that it's shipped in 1.26.0 stable?

[TyPR124]

Is there still intent to make the debug_upper_hex and debug_lower_hex flags public? See:

rust/src/libcore/fmt/mod.rs

Lines 1743 to 1751 in ef92009

	// FIXME: Decide what public API we want for these two flags.
	// https://github.com/rust-lang/rust/issues/48584
	fn debug_lower_hex(&self) -> bool {
	self.flags & (1 << FlagV1::DebugLowerHex as u32) != 0
	}
	fn debug_upper_hex(&self) -> bool {
	self.flags & (1 << FlagV1::DebugUpperHex as u32) != 0
	}

I would find these useful.

[Kimundi]

I just "rediscovered" these {:x?} formatting options. A short summary of the history here:

• Hexadecimal integers with fmt::Debug, including within larger types rfcs#2226 proposed this feature, including:
  • grammar changes to the formatting syntax to account for the new radix position where the x gets inserted in {:x?}
  • new fmt::Formatter APIs for querying that value
  • extension of the Debug impl for integers to conditionally delegate to fmt::LowerHex or fmt::UpperHex if this is set.
• Discussion on the original RFC thread lead to open questions about how to design the Formatter API in a way that is future-proof with possible other extensions like this.
• As reaction to that, it got implemented in Add hexadecimal formatting of integers with fmt::Debug #48978 without public Formatter API changes.

However, as a result of this the discoverability of this feature is kinda bad, as the only way to know about it is to read the two lines in std::fmt's docs that talk about it. Otherwise it's completely internal magic hidden between the Formatter type and Debug impls for integers, that can not be externally used for something else, and is thus completley invisible in the API.

This is also confusing because the syntax section in std::fmt's docs never got updated with the new grammar.