strconv: Unquote inconsistently decodes invalid UTF-8

[bjkail]

What version of Go are you using (go version)?

Playground.

What did you do?

strconv.Unquote with a quoted string containing an invalid UTF-8 byte sequence and either an escape sequence or not: https://play.golang.org/p/VZUxFxt7gh9

What did you expect to see?

The same encoding of the invalid UTF-8 regardless of whether the string contains an escape sequence (probably a reencoding of RuneError in both cases), or documentation explaining there's a difference.

What did you see instead?

If the string does not contain an escape sequence, the invalid UTF-8 sequence is preserved.
If the string contains an escape sequence, the invalid UTF-8 sequence reencodes RuneError.

The trivial code path does not validate the UTF-8.

[gopherbot]

Change https://golang.org/cl/92535 mentions this issue: strconv: detect invalid UTF-8 in the Unquote fast path