I wonder if both are needed. recordAccept for a MIME type (possibilities 
include application/xml, or application/json, or application/rdf+xml)

And recordSchema which says something like 'bibframe' or 'marc'.  Record 
schema like 'bibframe' or 'marc' could be 'registered' simply on a wiki 
page controlled by the 'SRU standards body'.  (If I understand right 
that's currently how recordSchema is used/intended?)

While there are still theoretical ontological complexities -- (if you 
want marcxml, is that recordAccept application/xml with recordSchema 
marc, or recordAccept application/marc+xml with recordSchema marcxml, or 
some other combination) -- I suspect in practice it would just Work Out, 
as conventions are established for common patterns. (Trying to specify 
this all absolutely with an elegant theoretical framework ends up just 
over-complicating things and not actually leading to increased 
inter-operability anyway, in my experience).

On 2/20/14 9:49 AM, Ray Denenberg wrote:
> No, that misses the point of recordSchema.  An SRU server might have a
> bibliographic database and be able to convert records to and supply them
> in  mods, bibframe, and bibframex, where the latter two are both rdf.
> recordSchema on the request says  which one is desired.  Just saying
> rdf, or rdf and json, isn't sufficient.
>
> Ray
>
>  > -----Original Message-----
>
>  > From: SRU (Search and Retrieve Via URL) Implementors
>
>  > [mailto:[log in to unmask]] On Behalf Of Jörg Prante
>
>  > Sent: Thursday, February 20, 2014 3:40 AM
>
>  > To: [log in to unmask]
>
>  > Subject: Antw: Re: SRU recordPacking and JSON
>
>  >
>
>  > There is no need for a Bibframe experimental media type. Bibframe model
>
>  > is a subset of RDF. So we already have
>
>  >
>
>  > application/rdf+xml
>
>  > application/ld+json
>
>  > text/n3
>
>  > text/turtle
>