Re: [ietf-types] Registration of media type application/senml+json

Ned Freed <ned.freed@mrochek.com> Wed, 22 September 2010 01:31 UTC

Message-id: <01NS5SX4C0CK000CVX@mauve.mrochek.com>
Date: Tue, 21 Sep 2010 18:15:43 -0700
From: Ned Freed <ned.freed@mrochek.com>
In-reply-to: "Your message dated Tue, 21 Sep 2010 16:00:44 -0600" <4C992B0C.7000407@stpeter.im>
MIME-version: 1.0
Content-type: TEXT/PLAIN
References: <FFD51C60-9887-47E4-B207-B5BC1AFBD67E@cisco.com> <4C992B0C.7000407@stpeter.im>
To: Peter Saint-Andre <stpeter@stpeter.im>
Cc: ietf-types@ietf.org
Subject: Re: [ietf-types] Registration of media type application/senml+json
Precedence: list

> On 9/21/10 1:12 PM, Cullen Jennings wrote:

> >   Type name:  application
> >
> >    Subtype name:  senml+json

> I thought that the "+xml" convention was for the special class of media
> types that represent XML MIME entities (RFC 3023). Is there a similar
> class of "JSON MIME entities" that would justify "+json"? I don't see a
> definition for such a class in RFC 4627 (not that I think it's a bad idea).

The situation surrounding the use of +suffix constructs is a little unclear.
RFC 3023 set up the rules for +xml fairly well (but see my earlier comments
about possible coupling of these suffixes to parameters like charset). However,
RFC 3023 said nothing about other +suffix constructs. And people definitely
want to use the convention for other popular structuring syntaxes. And all RFC
4288 has to say is:

    More generally, "+suffix" constructs should be used with care, given the
    possibility of conflicts with future suffix definitions. 

In my role as IANA media tyypes reviewer, I have interpreted this as meaning
that the +suffix construct should only be used when "suffix" actually
identifies a well known structuring syntax. I have pushed back on other
uses (I believe there was someone who wanted to register something like
application/vnd.typename+vendorname, for instance). But beyond that I have
allowed such registrations to go through.

What I do not do, and wish I could do, is push back on registrations of types
based on, say, JSON, saying they should be using +json. I don't believe the
current rules allow me to do that.

What I think is needed here is a revision/update to RFC 4288 to establish a
registry of allowed +suffix constructs as well as expanded guidelines for their
use. I also think pretty much any well defined structuring syntax should be
allowed in the registry. (It's not like there's a shortage of them either.)

				Ned

[ietf-types] Registration of media type applicati… Cullen Jennings
Re: [ietf-types] Registration of media type appli… Bjoern Hoehrmann
Re: [ietf-types] Registration of media type appli… Peter Saint-Andre
Re: [ietf-types] Registration of media type appli… Bjoern Hoehrmann
Re: [ietf-types] Registration of media type appli… Ned Freed
Re: [ietf-types] Registration of media type appli… Peter Saint-Andre
Re: [ietf-types] Registration of media type appli… Paul Libbrecht
Re: [ietf-types] Registration of media type appli… Ned Freed
Re: [ietf-types] Registration of media type appli… Keith Moore
Re: [ietf-types] Registration of media type appli… Mario Salzer
Re: [ietf-types] Registration of media type appli… Ned Freed
Re: [ietf-types] Registration of media type appli… Cullen Jennings