Consistency Namespace vs. ClosedNamespace

[mthag]

As I understand rdflib.Namespace and ClosedNamespace, they are both meant as auxiliaries to simplify the creation of URIRefs in Python code, e.g. RDF.type or OWL.Class. In addition, ClosedNamespace provides an additional vocabulary check, e.g., RDF.spam will raise an exception.

However, sometimes other uses of Namespace/ClosedNamespace can be useful, e.g., string comparison. At this stage, it is quite inconvenient that Namespace is derived from unicode, whereas ClosedNamespace is derived from the standard object, and the actual namespace string is an instance attribute (uri). In the interest of consistency, I propose that ClosedNamespace should also be derived from unicode (or even from Namespace).

Regards

[joernhees]

can you give us a use-case where you can't just compare str(Namespace(...)) with str(ClosedNamespace(...))?

[gromgull]

I totally agree that it's weird that they derive from different things. It also looks trivial to fix. I've tried - let me see if Travis agrees :)

[niklasl]

Also totally agree, thanks for addressing this! 👍

[joernhees]

the change mentioned in #551 is backwards incompatible and will be moved to rdflib 5.0.0 .