go/scanner: improve error message when a Go file is encoded as UCS-2

[adonovan]

When a Windows user mistakenly creates a Go text file encoded as UCS-2 (i.e. UTF-16 2-byte encoding) the error message they see is "unexpected NUL", from the zero byte in the encoding of the first rune of the file (typically one of [ \n/p]). The scanner could issue a more helpful error: "encoding of file foo.go is UCS-2; Go requires UTF-8".

(The presence of a U+FEFF BOM is not by itself indicative of a UCS-2 encoding.)

See also golang/vscode-go#3690

[gabyhelp]

Related Issues

• cmd/go: detect UTF-16 encoded sources and improve error reporting #71872

_{(Emoji vote if this was helpful or unhelpful; more detailed feedback welcome in this discussion.)}

[ianlancetaylor]

See also #71872 .

[gopherbot]

Change https://go.dev/cl/652515 mentions this issue: go/scanner: report specific error for UCS-2 encoded files