fmt shouldn't touch up hex escape sequences in zon and extern identifiers (`.@"\x0a"` => `.@"\n"`)

[SeriousBusiness101]

Input:

.{
                      .@"\x01" = .{""}, .@"\x02" = .{""}, .@"\x03" = .{""},
    .@"\x04" = .{""}, .@"\x05" = .{""}, .@"\x06" = .{""}, .@"\x07" = .{""},
    .@"\x08" = .{""}, .@"\x09" = .{""}, .@"\x0A" = .{""}, .@"\x0B" = .{""},
}

Currently fmt outputs:

.{
    .@"\x01" = .{""},
    .@"\x02" = .{""},
    .@"\x03" = .{""},
    .@"\x04" = .{""},
    .@"\x05" = .{""},
    .@"\x06" = .{""},
    .@"\x07" = .{""},
    .@"\x08" = .{""},
    .@"\t" = .{""},
    .@"\n" = .{""},
    .@"\x0b" = .{""},
}

Instead of:

.{
    .@"\x01" = .{""},
    .@"\x02" = .{""},
    .@"\x03" = .{""},
    .@"\x04" = .{""},
    .@"\x05" = .{""},
    .@"\x06" = .{""},
    .@"\x07" = .{""},
    .@"\x08" = .{""},
    .@"\x09" = .{""},
    .@"\x0a" = .{""},
    .@"\x0b" = .{""},
}

Edit:

Most relevant for ZON and extern. Should probably only apply to those? What I'm suggesting is fmt shouldn't perform this canonicalization on interchangable bytes representations where they're known to exchange out of Zig's realm.

[castholm]

Is there a strong motivation for this beyond a desire to align source code? Keep in mind that you can disable the auto-canonicalization with // zig fmt: off. The original motivation for canonicalizing identifiers was greppability; see #166 and its linked pull request. ASCII bytes in identifiers currently have only one canonical representation each.

[SeriousBusiness101]

@castholm The motivation as in my other issues is foremostly object notation. The desire is 8-bit cleanness, in situations where you're reaching outside of a vacuum. I don't think it serves well to arbitrarily canonicalize interchangable byte representations, at the crossroads where they're interchanged. Andrew's words https://github.com/ziglang/zig/issues/14534:

Even JSON has the ability to represent null bytes in map keys. It would be a crime to introduce a new data exchange format and have it be worse than JSON at representing a mere map data structure.

[silversquirl]

Not really sure how Andrew's comment there is relevant here. Canonicalizing helps with searchability, as @castholm says, and if you really want to disable it you can turn off zig fmt for those lines

[SeriousBusiness101]

Actually this form of canonicalization isn't a measure that helps with searchability, most certainly not in the example I presented here @silversquirl. Andrew knows a crime when he sees one.

[silversquirl]

canonicalizing helps with searchability in the common case because it means you can search for @"\t" and find all the fields that are named a single tab character. Your usecase (non-textual field names) is uncommon, and not one that I think Zig should design around.

[SeriousBusiness101]

@silversquirl It's not a non-textual use case. It's disengaged from an exclusive canonicalization for a specific encoding, in what is essentially it's own interchange format. This makes sense in extern too. There isn't profound complexity to grep conditions, and I don't think this friction in particular is bad for that matter.