[occi-wg] OCCI Categories and Types
andy.edmonds at gmail.com
Fri Aug 13 08:26:05 CDT 2010
On 13 Aug 2010, at 13:34, Ralf Nyren wrote:
> On Fri, 13 Aug 2010 13:36:56 +0200, Alexander Papaspyrou
> <alexander.papaspyrou at tu-dortmund.de> wrote:
>> I don't recall that this ever has been discussed, but it certainly is a
>> very good idea.
> The old (svn r164) docbook spec said:
> "Attributes defined by this standard reside at the root but anyone can
> define a new attribute by allocating a unique namespace based on their
> reversed Internet domain (e.g. “com.example.attribute”).
> I think such a namespace policy for attributes would be useful for the
> examples in the spec and related documents as well. An example is much
> easier to understand if the contents is well defined.
AE: yep I agree with all of this - @Ralf would you be willing to place this up on the wiki under a page named something along the lines of "Extensions"
>>> I do not quite understand the various Content-type headers used in the
>>> examples. Are they of any significance to the definition/use of
>> No. They are just supposed to make it easier to spot the type within the
>> core model. So, the MIME types would be something like
>> application/occi-resource (for the Resource class from core)
>> application/occi-link (for the Link class from core)
>> ... (whatever else is defined as a class in core)
>> That way, you don't have to analyze the details of a REST resource, but
>> just look at the MIME type delivered by the OCCI container.
> If just for the purpose of the example I can somewhat agree. Otherwise I
> would say the Content-type header only reflect the body and not what kind
> of information you happen to have in the header.
IMO, header and body are explicitly linked by content type - e.g. HTML - content type is specified as HTML in the header and the body will reflect this. It doesn't make sense to specify HTML as the content type in the header and then send a JPEG ;-)
>>> - In Link header: rel="category http://prov.com/occi#action"
>>> Is the format rel="category XXX" something new as well?
>> Yes. We tried to stick as much as possible to the IETF nottingham Link
>> draft , not adding any additional vendor-specific extensions.
> Hmm... reading the RFC-to-be version (from August 2010) I do not find
> "category" to be a registered relation type. Defined in some other
"category" is not a registered rel type but can be. Do note that what is in the Category wiki page are merely explorations at this stage and so hence some vague aspects.
> regards, Ralf
> occi-wg mailing list
> occi-wg at ogf.org
More information about the occi-wg