8-bit \x escapes

[mattheww]

Problem 1

The definition of ByteEscape in §2.4.1 (fls_2ifjqwnw03ms) includes \x OctalDigit HexadecimalDigit.

ByteEscape is then used in the definition of byte literals (also in §2.4.1) and simple byte string literals (in §2.4.2.1, fls_msbaxfc09vkk).

But byte literals and simple byte string literals can also include 8-bit \x escapes (that is, escapes in which the first character after the \x is a HexadecimalDigit which isn't an OctalDigit).

So the FLS is mistakenly saying that, for example, b'\xff' isn't accepted as a byte literal and b"\xff" isn't accepted as a simple byte string literal.

Similarly the definition of SimpleCStringLiteral is based on AsciiEscape, which again includes \x OctalDigit HexadecimalDigit, but 8-bit \x escapes are allowed in simple c string literals.

Problem 2

§2.4.2.1:2 (fls_moe3zfx39ox2) reads:

A simple byte string literal is a byte string literal that consists of multiple AsciiCharacters.

while AsciiCharacter is defined in §2.1:16 (fls_PIDKEm8GiLNL) as follows:

An AsciiCharacter is any Unicode character in the range 0x00 - 0x7F, both inclusive.

But (because of the 8-bit \x escapes described above) a simple byte string literal can contain arbitrary bytes.

So I think §2.4.2.1:2 should say that a simple byte string literal is made up of bytes (compare §2.4.1:2 (fls_q0qwr83frszx) "A byte literal is a literal that denotes a fixed byte value.")

Also, where §2.4.2.1:3 (fls_vffxb6arj9jf) says

The type of a simple byte string literal of size N is &'static [u8; N].

I think it would be good to be more explicit than "size".