Questions in case of multipul Types for one Service

[OHTSUKA Ko-hei]

OK, thank you, I 'maybe';) understand.

Reason I asked is, not thinking it is invalid or misleading, but just
want to know how to handle it.

These days, Yadis service proveiders evolved day by day to implement
new specification (like, MyOpenID supports Simple Registration).
It is good, but for delegating users, little painful to update
delegating Yadis document, by surveying our all servers' capability
manually.
   'Does MyOpenID implement sreg extension or not?'
   'How about Videntity.org?'
   'How about LiveJ..'
Painful.

So, I start to make little tool for automatic update own delegating
Yadis document, by visiting 'delegated' Yadis URL at periodic intervals.
If 'delegated' Yadis capabilities are updated, the tool also updates
local Yadis document.

But, I wonder for such case:
My own Yadis document is:

<Service priority='0'>
  <Type>http://lid.netmesh.org/minimum-lid/2.0b8</Type>
  <URI>http://mylid.net/foo</URI>
</Service>
<Service priority='1'>
  <Type>http://lid.netmesh.org/sso/2.0b8</Type>
  <URI>http://mylid.net/foo</URI>
</Service>

and, when the tool visit http://mylid.net/foo, it found that
http://mylid.net/foo implements new extension Type.
   http://lid.netmesh.org/extreme-special-extension/2.0 ;)
In such case, which service elements (priority 0 or 1) to add
new type?
Or, should make new service element? then, use how much priority?

That is my wondering problem, so ask questions on former mail.

Johannes Ernst wrote:
> Your question is, I think, about the NetMesh implementation, so let  me 
> try and answer it.
> 
> My understanding of what we defined in Yadis is that:
> 
> <Service>
>  <Type>A</Type>
>  <Type>B</Type>
>  <URI>U</URI>
> </Service>
> 
> is semantically equivalent to
> 
> <Service>
>  <Type>A</Type>
>  <URI>U</URI>
> </Service>
> <Service>
>  <Type>B</Type>
>  <URI>U</URI>
> </Service>
> 
> At least for Types A and B that are not too tightly dependent on each  
> other. So far, anybody disagree with that?
> 
> You are right that there is no point in assigning different  priorities 
> to "non-substitutable" Service types. ("Non-substitutable"  meaning here 
> that I cannot use format negotiation instead of VCard  queries, while 
> "substitutable" would be LID GPG-based SSO vs. OpenID  Sign-on)
> 
> But then, I don't think it is invalid or misleading either: a  consumer 
> of this file will simply look for the Service Types it  understands, and 
> of those the kinds it is looking for, and find 0 or  1 matches, thereby 
> ignoring priorities (with the exception of LID GPG- based SSO vs. OpenID 
> SSO). So I don't think the file we produce is  either wrong or misleading.
> 
> (But the real reason that you are seeing this is that our simple  
> implementation behind it represents the set of Services at this URL  in 
> an array, ordered by priority, and when it dumps it out, it just  
> assigns the array index as priority, without looking at  
> substitutability, and without looking whether or not they could be  
> "collapsed" into a single service statement).
> 
> If somebody makes a convincing argument why the latter is needed,  we'll 
> implement it ...
> 
> 
> On Apr 14, 2006, at 16:04, OHTSUKA Ko-hei wrote:
> 
>> Question.
>>
>> On OpenID, extensional protocol shows in Yadis,
>>
>> <Service>
>>   <Type>http://openid.net/signon/1.2</Type>
>>   <Type>http://openid.net/extensions/sreg/1.0</Type>
>>   <URI>http://www.myopenid.com/server</URI>
>> </Service>
>>
>> But on LID, many Yadis example shows like
>>
>>   <xrd:Service priority="1">
>>    <xrd:Type>http://lid.netmesh.org/minimum-lid/2.0b8</xrd:Type>
>>    <xrd:URI>http://mylid.net/foo</xrd:URI>
>>   </xrd:Service>
>>   <xrd:Service priority="2">
>>    <xrd:Type>http://lid.netmesh.org/sso/2.0b8</xrd:Type>
>>    <xrd:URI>http://mylid.net/foo</xrd:URI>
>>   </xrd:Service>
>>   <xrd:Service priority="3">
>>    <xrd:Type>http://lid.netmesh.org/sso/1.0</xrd:Type>
>>    <xrd:URI>http://mylid.net/foo</xrd:URI>
>>   </xrd:Service>
>>   <xrd:Service priority="4">
>>    <xrd:Type>http://lid.netmesh.org/relying-party/2.0b8</xrd:Type>
>>    <xrd:URI>http://mylid.net/foo</xrd:URI>
>>   </xrd:Service>
>>   <xrd:Service priority="5">
>>    <xrd:Type>http://lid.netmesh.org/traversal/2.0b8</xrd:Type>
>>    <xrd:URI>http://mylid.net/foo</xrd:URI>
>>   </xrd:Service>
>>
>> So, I have two question.
>>
>> 1. If there are no priority differences among every services,
>>    upper xrds meaning may be same as
>>
>>   <xrd:Service>
>>    <xrd:Type>http://lid.netmesh.org/minimum-lid/2.0b8</xrd:Type>
>>    <xrd:Type>http://lid.netmesh.org/sso/2.0b8</xrd:Type>
>>    <xrd:Type>http://lid.netmesh.org/sso/1.0</xrd:Type>
>>    <xrd:Type>http://lid.netmesh.org/relying-party/2.0b8</xrd:Type>
>>    <xrd:Type>http://lid.netmesh.org/traversal/2.0b8</xrd:Type>
>>    <xrd:URI>http://mylid.net/foo</xrd:URI>
>>   </xrd:Service>
>>
>>    Is this understanding right?
>>
>> 2. I have little knowledge about LID, so writing below is maybe
>>    misunderstanding,
>>    But I think LID's each protocols (sso,relying-party,traversal,
>>    etc...) are used for different usecases, so adding different
>>    priorities to them is nonsense.
>>
>>    Though seeing XRD files examples including LID, almost all
>>    examlpes adding different priority to each LID protocol.
>>
>>    Are there any reason to do like this?
> 
> 
> Johannes Ernst
> NetMesh Inc.
> 
> 
> ------------------------------------------------------------------------
> 
>  http://netmesh.info/jernst
> 
> 
> 
>