An empty signature will match all things

[ross-spencer]

Given the signature file:

<?xml version="1.0" encoding="UTF-8"?>
<FFSignatureFile xmlns="http://www.nationalarchives.gov.uk/pronom/SignatureFile" Version="1" DateCreated="2024-03-28T14:47:56+00:00">
  <InternalSignatureCollection>
  <InternalSignature ID="1" Specificity="Specific">
  <ByteSequence Reference="BOFoffset" />
</InternalSignature>
</InternalSignatureCollection>
<FileFormatCollection>
  <FileFormat ID="1" Name="ACME Generic Sequence of Bytes" PUID="ACME/1" Version="2600" MIMEType="x-application/acme">
    <InternalSignatureID>1</InternalSignatureID>
    <Extension>acme</Extension>
  </FileFormat>
</FileFormatCollection></FFSignatureFile>

When loaded into DROID and run against an arbitrary set of objects, DROID will return a signature identification for all objects.

I am not sure the impact on this in a larger dataset, and it may be unlikely that PRONOM will output a signature with this problem, but from a digital preservation standpoint it feels wrong and could cause some potential issues for folks.

[nishihatapalmer]

An interesting edge case - from a purely mathematical standpoint, the empty string matches everything! I'm actually surprised that this does match everything though, didn't think it was possible to compile a matcher with no content, thought that would generate an exception.

Even so, if this ever happened, it would be obvious that there was a problem with the signature, and the signature would be fixed. Not sure this is worth spending any time fixing in the code.