I received a bug report on my skeleton-test-suite generator recently. The issue known, surrounding the placement of the two byte-sequences in fmt/189: http://www.nationalarchives.gov.uk/pronom/fmt/189
In short, I wonder if the PRONOM team could consider at a minimum, swapping the two byte sequences around, by re-writing them in sequence in PRONOM, and/or replacing the two sequences for one.
My initial thinking is that this would have equivalent functionality:
Position type Absolute from BOF
Offset 0
Byte-order Little-endian
Value 504B0304{0-18}504B0102{0-18}504B0506{0-18}5B436F6E74656E745F54797065735D2E786D6C20A2
To the original:
Sequence one:
Position type Absolute from BOF
Offset 30
Byte order Little-endian
Value 5B436F6E74656E745F54797065735D2E786D6C20A2
Sequence two:
Position type Absolute from BOF
Offset 0
Byte order Little-endian
Value 504B0304*504B0102*504B0506
My working might be a bit rough, an approximation. A variation on the sequence above might certainly satisfy the requirement.
The benefit might be small in this instance so it is appreciated that it wouldn't be top priority.