Clarify the rules around PrefixDecl in the grammar where PNAME_NS not unique

[JervenBolleman]

Some implementations do not allow multiple identical prefix declarations, others do

prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>

fails in some stores, the behavior of

prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
prefix rdfs: <wrong:wrong>

Is implicit, but not explicitly stated, to take the last definition of the prefix rdfs:.

Would a pull-request stating that the rules are the last of the duplicate prefix declaration should be chosen be welcome?

[afs]

How many implementations?

In RDF 1.2 Turtle, https://www.w3.org/TR/rdf12-turtle/#namespaces, the namespaces state is a map and the process is in the order of parsing.

The section https://www.w3.org/TR/rdf12-turtle/#sec-iri :

Subsequent @prefix or PREFIX directives may re-map the same prefix label.

That text could be a short note with a mention of "same re-map".

See also these issues raised on 2025-02-13:
w3c/sparql-update#54
w3c/data-shapes#239
w3c/rdf-turtle#87

[JervenBolleman]

Subsequent @prefix or PREFIX directives may re-map the same prefix label.

That text could be a short note with a mention of "same re-map".

Yes, that is sufficient and I would be happy to draft a test case as well.
Perhaps the same for the BASE keyword

[afs]

BASE is interesting because the parsing rules don't distinguish the places IRIs are used in the document and apply to directives:

BASE <http://example/>
BASE <x>

is legal.

As is

BASE <http://example/>
PREFIX ns: <namespace>