IRI fragments cannot start with a digit

[twalmsley]

I'm using random UUIDs after the # in IRIs and some of them start with digits, but the prefix replacement regex doesn't match these IRIs. Is there a reason for that or is it a bug? I can't find anything in the W3C that restricts fragments in this way.

E.g. https://www.example.com/test/things#3add0a26-8c75-47fb-abe7-a0ed1c23a0a doesn't match the generated regex:
^(?:hqdm:|amrc:|rdf:)[^/]*$|^(https:\/\/www\.example\.com\/test\/things#)([_a-zA-Z][\-_a-zA-Z0-9]*)$

The result is that some IRIs are not shortened using a prefix when exported (to TLL in my case), although the TTL is still valid.

[RubenVerborgh]

That's by design indeed; the Writer prefixes some URIs, but not all of them. The regex can be expanded indeed, but getting it perfect for all cases would make it slow. That said, yours is a reasonable case that I wouldn't expect to slow down things.