Multiple prefix declarations in SPARQL query allowed or not?

[JervenBolleman]

e.g. we currently throw errors on

prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
SELECT *{
  ?x rdfs:label ?y
}

should we allow these? and what should we do in the case of

prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
prefix rdfs: <wrong:wrong>
SELECT *{
  ?x rdfs:label ?y
}

Other SPARQL endpoints allow these.

[hmottestad]

Maybe we could allow it for the extended SPARQL evaluation. Mind you, the SHACLE engine needs to use the strict version of SPARQL evaluation to comply with the spec.

[JervenBolleman]

The SPARQL spec is not clear about this w3c/rdf-star-wg#160 i think the intent was to allow it like turtle.

[hmottestad]

https://www.w3.org/TR/sparql11-query/#iriRefs

Base IRIs declared with the BASE keyword must be absolute IRIs. A prefix declared with the PREFIX keyword may not be re-declared in the same query.

[hmottestad]

I actually couldn't find it myself, but I got the research feature in ChatGPT to work on it found the sentence in spec.

[nguyenm100]

not to be pedantic but does redeclared mean redefined (eg different l)?

[hmottestad]

Don't know actually. If you re-declared the same prefix and namespace it probably shouldn't be a problem. If you do the same for a relative namespace then I think it would be different if it is declared before or after @base, but if the resolved namespace is the same the same then it would be a pure duplicate and maybe also not considered a re-declaration.

[frensjan]

I voted Disallowed (like now) with the voice-over that I'd ideally would like a fourth option: redefining prefixes should throw an error as in you second example, while redundant / duplicate definitions should be accepted (ignored).

[frensjan]

That's also my interpretation of what @hmottestad quoted from the spec:

A prefix declared with the PREFIX keyword may not be re-declared in the same query.