Re: section 4.1.5

> > 4.1.5 "LDPRs MUST use the predicate rdf:type to represent the 
> > concept of type."
> > 
> > Why does this have to be a MUST ? 
> > I don't think that is required to ensure that anything else works.
> > I would say this is a SHOULD, or a best practice .. 
> > ?

No objection to issue-ing it.

Few things come to mind as background on the origin of this.  I leave it 
to the WG for now to decide which is best in normative vs informative 
materials.

1: Some people started using dcterms:type, and we wanted LDP clients to 
have a single reliable (always there) way to know of an RDF resource was 
of the right type.  While Dublin Core has words to discourage that for RDF 
resources (I know because I searched for an found them), there are many 
paths into Dublin Core and frankly the words are something you don't find 
unless you're either looking for them (most won't know to do so) or you're 
studying DC itself in depth (and honestly "no one" does this).  I tell all 
our devs they have to read HTTP too, but I know they don't and I 
sympathize - time is a commodity and we all choose where to spend it, not 
always wisely in hindsight.

2: (this one is almost certainly deployment rather than normative) Some 
clients were [foolishly I know, but it happened] processing RDF/XML using 
non-RDF-aware XML processors and coding their processing to be dependent 
on abbreviated syntax (so root element qname, transformed to a URI == 
rdf:type URI).  They did not fare well with servers that used 
rdf:Description root elements and hence "moved the type" to rdf:type.  Nor 
would those clients behave reliably with >1 rdf:type.  Others wrote code 
to assume rdf:Description-style and broke when abbreviated syntax was 
given them.  Yes we tell them to use RDF libraries, but I'll tell you from 
personal experience not all products have that within their adoption cost 
envelope.  I know of at least one product still using its own hand-crafted 
HTTP server (from pre-Apache days).




Best Regards, John

Voice US 845-435-9470  BluePages
Tivoli OSLC Lead - Show me the Scenario

Received on Friday, 31 May 2013 18:51:54 UTC