Does sh:uniqueLang need to be updated for rdf:dirLangString

[HolgerKnublauch]

I am not really familiar with rdf:dirLangString, but the current definition of sh:uniqueLang

https://www.w3.org/TR/shacl12-core/#UniqueLangConstraintComponent

refers to "language tags" only. While the text direction may be stored in the language tag in TTL syntax, see

https://www.w3.org/TR/rdf12-turtle/#ex-quoted-literals

I am not clear whether our use of language tags matches the RDF 1.2 data model spec

https://www.w3.org/TR/rdf12-concepts/#dfn-dir-lang-string

(There may be other places beside sh:uniqueLang where this is relevant).

I guess at minimum, we should add a test case to clarify whether "A"@ar--rtl is the same as "A"@ar--ltr or not.

[afs]

The text direction is not part of the language tag.

Literals are now 2, 3 or 4 parts -- lexical form, datatype, (langtag (, textdirection)?)?

In Turtle, it is convenient to place text direction after the language tag because the language tag is required if text direction is used. But it is a separate component of a literal.

In RDF/XML, it a separate XML attribute its:dir.

[HolgerKnublauch]

Yes, that is the definition in RDF 1.2.

My question now is: should A"@ar--rtl and "A"@ar--ltr count as distinct (for sh:uniqueLang) or not. I.e. is a subject that has both values violating a sh:uniqueLang constraint? I don't know.

[afs]

For the UC of having at most one displayable item of text, they are different WRT sh:uniqueLang. c.f. @fr, @fr-ca.

Are there other UCs?

[HolgerKnublauch]

Ok, the PR #750 changes sh:uniqueLang so that different text direction counts as different language tag.

[afs]

Ok, the PR #750 changes sh:uniqueLang so that different text direction counts as different language tag.

If you want a principle, then same style as the condition applied to @en, @en-gb, @en-us.

It seems that uniqueLang (even SHACL 1.0) is not "one displayable text fragment UC. @en, @en-gb, @en-us are different language tags.