Re: [ietf-types] Registration of media type application/opensearchdescription+xml

[Frank Ellermann <[e-mail]>]

On 25 October 2011 08:54, Paul Libbrecht wrote:

> I hope someone will review the formal aspects of the
> procedure.

RFC 4288 guarantees an IETF Last Call for this beast if
it ever gets this far.  I've tried to submit an updated
"not for RFC" Internet Draft, but that's still stuck in
a manual submission queue.

RFC 4288 also guarantees (or rather demands) a complete
copy of the registration template in whatever passes as
specification, RFC or otherwise:

<URL:http://www.opensearch.org/Specifications/OpenSearch/1.1/Draft_5#MIME_type_application.2Fopensearchdescription.2Bxml>

Content and/or URL might change during the review here.

The OpenSearch folks suggested to replace OSDD by OSD,
and to add ".osdx" as the conventional file extension:
Microsoft uses ".osdx".

I'm not interested to "invent" wild + wonderful features,
I'm only interested to get this beast registered "as is".

> My usual suggestion is: what about clipboards?

Dunno, is there any problem with XML and clipboards?

> OpenSearch Description Documents sound to be a natural
> candidate to be exchanged in clipboards (be copied-and
> -pasted, be drag-and-dropped, ...). Do I mistake?

All I ever wish to do with OSDs is (1) edit some details,
or wrt the security considerations inspect their source,
(2) adopt an OSD when the auto-detect link rel="search"
worked as it should, (3) click on an AddSearchProvider()
EcmaScript link for cases where link rel="search" is no
option.  Where do clipboards enter this picture for you,
and is it relevant for the IANA media type registration?

> - Macintosh Uniform Type Identifier:
> org.opensearch.description conforms to public.xml

Sadly I've no idea how Mac Uniform type identifiers work,
and the info published in RFC 4288 convinced me that it's
not something to worry about.  If you have any "official"
public source that this is a well-known Mac id. I add it,
otherwise I prefer to stick to what I could defend in an
IETF Last Call.

> My guts feelings about the "responsibility" part of the > registration is that it lacks a tiny bit of formality:
> - isn't the WhatWG HTML a "place" for drafting a format > that will hopefully end up be a W3C standard?

AddSearchProvider() is documented in WhatWG HTML.  Both
the HTML5 draft and WhatWG HTML have a reference to the
OpenSearch 1.1 spec.  The rel="search" link relation is
registered for about a year.  Oh yes, there is this tiny
bit of "happiana" and RFC 4288 standards tree, but "let's
see what happens" (before giving up on this registration).

> - a google-group as the contact is really volatile.

Well, it's how the OpenSearch community works, they have a
MediaWiki installation with talk pages, and a mailing list
hosted by Google groups.  For at least five years, I can't
tell how it was when this work was still hosted by A9.com.
AFAIK Amazon still donates the domain, and they certainly
donated the OpenSearch XML name space URI some years ago.

> Is there no way to put a legally registered entity
> instead? I think even a9.com would be better.

I'm trying to register a well-known media type used in all major
browsers and on many web pages, but I cannot claim
that I try this on behalf of A9.

The OpenSearch.org folks know that I want this media type
to be registered since I-D.ellermann-opensearch-00 (2008),
and so far nobody had objections.  Sadly there were no
volunteers to create an RFC, and the proposal to create a
DTD for a future RFC died in a resounding silence.

> - Similarly http://www.opensearch.org/ is rather a
> volatile change controller.

This actually is the "change controller" for some years,
but IANAL and certainly not speaking for A9.

-Frank

<URL:http://www.opensearch.org/User:Xyzzy>
<URL:http://www.opensearch.org/Specifications/License>