This is in regard to Sequence Alignment/Map Format Specification (2022-08-22).

§ 1.3 "The header section" defines patterns for header lines:

Thus header lines match /^@(HD|SQ|RG|PG)(\t[A-Za-z][A-Za-z0-9]:[ -~]+)+$/ or /^@CO\t.*/.

This invalidates the following test examples:

• test/sam/passed/hdr.PG3.sam

  @PG ID:bwa-1 CL:echo "🐟"

• test/sam/passed/hdr.PG5.sam

  @PG ID:x DS:☢ description ☢

• test/sam/passed/hdr.SQ6.sam

  @SQ SN:tea LN:99 DS:☕

The text "UTF-8 encoding may be used" for the CL and DS fields does not remove the character set constraint. It also remains arbitrary as to why only some fields have this definition.