You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@shindig.apache.org by Paul Lindner <pl...@hi5.com> on 2009/01/05 19:26:25 UTC

Re: Uhm, LinkSpec rel value?

On Dec 30, 2008, at 7:11 PM, Kevin Brown wrote:

> On Tue, Dec 30, 2008 at 3:58 PM, Henning Schmiedehausen <
> hgschmie@googlemail.com> wrote:
>
>> But the stuff in the gadget xsd is significantly different. This  
>> looks
>> more like "these are some defined like types for e.g. the icon of the
>> gadget, an URl that offers a help page and an URI to get support for
>> the gadget".
>
>
> I think that's the intent. I don't really know, since I always  
> thought the
> Link element was kind of lame and couldn't come up with a better  
> counter
> proposal so I stopped paying attention. Intellectual laziness.
>
> Now that you bring it up though, yeah, the XSD should probably just  
> specify
> this as a plain old string and let the prose of the spec explain  
> what it's
> used for.
>
> All the more reason to get things moving on
> http://wiki.opensocial.org/index.php?title=Prop_003:_Revised_Gadget_Specification:)
>
>

I proposed the <Link> tag to complement the xhtml link/rel tag used in  
html/atom.  The idea is to allow for a generic bag of URLs that can be  
purposed and repurposed as times goes on.

Ideally I'd like to see most opensocial objects have this concept  
embedded into the core.  The Person object has the 'urls' field, and I  
included it in the messaging API proposal.

I've already used it on hi5 to implement larger sized profile photos  
and plan to add other URLs as time goes on.

>>
>>
>> I finally got around to run an XSD editor / validator over the posted
>> specs. for both specs, they report
>>
>> src-attribute.4: Attribute 'rel' has both a 'type' attribute and an
>> anonymous 'simpleType' child. Only one of these is allowed for an
>> attribute.
>>
>> So in a nutshell, it seem that the whole block
>>
>> --- cut ---
>>                   <xs:simpleType>
>>                     <xs:restriction base="xs:string">
>>                       <xs:enumeration value="gadgets.help"/>
>>                       <xs:enumeration value="gadgets.support"/>
>>                       <xs:enumeration value="icon"/>
>>                     </xs:restriction>
>>                   </xs:simpleType>
>> --- cut ---
>>
>> is bogus. And should probably go from the spec definitions.
>
>
>
>
>>
>>   Ciao
>>        Henning
>>
>>
>> On Tue, Dec 30, 2008 at 13:41, Kevin Brown <et...@google.com> wrote:
>>> On Tue, Dec 30, 2008 at 10:53 AM, Henning Schmiedehausen <
>>> hgschmie@googlemail.com> wrote:
>>>
>>>> Hi,
>>>>
>>>> if I read the gadget xsd for 0.8 right, then the rel parameter of
>>>> <Link> element allows only three values: icon, gadget.help and
>>>> gadget.support:
>>>>
>>>> <xs:attribute name="rel" type="xs:string" use="required">
>>>> <xs:simpleType>
>>>> <xs:restriction base="xs:string">
>>>>  <xs:enumeration value="gadgets.help"/>
>>>>  <xs:enumeration value="gadgets.support"/>
>>>>  <xs:enumeration value="icon"/>
>>>> </xs:restriction>
>>>> </xs:simpleType>
>>>> </xs:attribute>
>>>>
>>>> However, various tests in LinkSpecTest on the 1.0.x release branch
>>>> happily use "foo" etc. values and actually do substitution on the  
>>>> rel
>>>> value (something which I haven't found anywhere in the spec at  
>>>> all).
>>>> Is that correct?
>>>
>>>
>>> There are actually an arbitrary number of values for rel to handle
>> lifecycle
>>> events, which can be container-defined.
>>>
>>> Oddly, there actually are some standardized event names, as listed  
>>> here:
>>>
>>>
>> http://www.opensocial.org/Technical-Resources/opensocial-spec-v081#TOC-Lifecycle-events
>>>
>>> But they're not in the XSD. I suspect this is just an oversight,  
>>> since
>> the
>>> lifecycle event proposal was kind of tacked on at the end of 0.8's
>> proposal
>>> cycle.
>>>
>>>
>>>>
>>>>
>>>>   Ciao
>>>>        Henning
>>>>
>>>
>>

Paul Lindner
plindner@hi5.com