RDF-star interest in profile-based content negociation

Hi all,

this message is long overdue, sorry for the delay.

I was supposed to explain why the upcoming RDF-star working group (1) 
would be interested in profile-based content negociation.

RDF-star extends various concrete syntaxes of RDF, for example Turtle. 
Old Turtle files are still valid (and have the same meaning) in the new 
version, but some new constructs are invalid in the old syntax, and 
would therefore break old parsers.

The defensive approach would be to define a new media-type for the new 
version of Turtle. Therefore, old consumers would not even try to parse 
new data). The problem is that up-to-date producers would probably 
expose the new media-type, even if they don't use (or rarely use) the 
new features.

The optimistic approach is to update the media-type, and expect all 
consumers to eventually upgrade -- counting the fact that, in the 
meantime, they will only rarely be exposed to new features. And when 
they do, they need to deal with the error anyway.

An intermediary approach is to update the media-type *with a notion of 
profile*. A provider could therefore expose content as 
text/turtle?profile=new or text/turtle?profile=old. To start with, this 
is the same as the optimistic approach: old consumers will ignore the 
profile, and might accidentally be exposed to new content. However, they 
could (relatively) easily be made "profile-aware", without requiring to 
upgrade them to support the new features (they would then explicitly 
require ?profile=old).

Does it make sense?


(1) RDF-star is currently a community group report, but a working group 
charter is in the work

Received on Tuesday, 24 May 2022 17:00:11 UTC