[OpenID] Open Challenge to webfinger and XRD

Peter Williams home_pw at msn.com
Sat Oct 24 19:44:27 UTC 2009 

In the debate below, 2 disctinct documents seem to be being discussed. One
document is an XRD (of some profile). The second document is also an XRD (of
another profile, say). Both documents are available on a single website, and
each can be referenced by URL(s). Folks may be arguing about the form of the
URL(s) locating the two document streams - since either might be retrieved.

Why not retain compatibility with the W3C identifier model, and combine it
with some of the excellent techniques used in the older XRD spec (XRI era)?

Why not only ever expect a website to expose 1 URL pointing to a particular
XRD document on an absolute URL (considering no fragments, as normal).
Having retrieved the XRD bytestream, now consider the fragment tag(s) - and
expect a tag to ref a particular Link's identifier within the stream (e.g.
the link's subject/localid field). Once de-referenced, one will determine
what kind of link it is (from the link's rel field, rather than the tag
value). In some cases, the rel field will indicate the *profile* of the
child XRD rather than more conventional uses of the rel typing on such has
HTML meta links. This "special" linking is akin to the special
nameresolution SEPs of the XRI era (which carved out a special role for
those SEPs that instrumented the resolver process "itself")

Now let each of the 2 Links (in one document) be a pointer to a separate XRD
document (one profile per link). These 2 documents, pointed to from the
"parent" by rel-typed links in the parent's XML, may be on the same website
as the parent XML (or not). If they are, they are on a different path to the
parent (and each other).

In the old XRI identifier world (that define how sequences of XRDs were
"valid"ated), we had no problem constructing sequences of XRD in which the
relationship of elements in the sequence was a function of compound name
resolution (or fancy redirect/referall processes that addressed high
availablity and delegation of authority "special" use cases). And, in that
the world the parent XRD's SEP that identified the next element in the
sequence had type declaring that said handoff introduced the child "in the
context of a parent". (In the trusted resolution variety of that handoff,
keying material borne in the parent SEP was crucial to establishing validity
of the child authorities trustworthiness (and its crypto signature) -
something that seems to be being carried over into the XRD 1.0 world).


Surely, in the "updated" metamodel of XRD 1.0, one could standardize two rel
URIs in 2 profile documents (much as one is apparently standardized "type"
URI), which when typing links can identify that a "subordinate" XRD document
is "how to discover the LRDD doc related to its parent XRD on a path" or
"how to now further discover the Personal XRD related to its parent XRD on
the path", or "how to further discovery the site-meta XRD related to its
parent on the path".

Now this sounds intuitive, feels like the older model, and aligns with
semweb ways of doing using fragment to locate a part of an XHTML document
(such as foaf files). But... perhaps the suggestion founders, as I can see
how its also now breaking the central goal of the moden OASIS work -
unbundle identifiers from XRD  format. After all, what did I just do? but
tie during processing the structure of identifier components (now path
elements vs XRI elements) to the processing of sequences of XRD.

I feel its legitmate for folks to discussing this on openid (even without a
workgroup focus). We know the OP vendors are doing domain-delegation via
proprietary means today (and this issue is not going away), and relating one
XRD to another (for the purposes of domain delegation using a Google like
discovery cloud) is assumed to be what will become a standard means of
accomplishing that which folks are doing today by proprietary means.

...



John Bradley-9 wrote:
> 
> The XRD may not be controlled by the same individual.
> 
> I understand your argument, and I understand LRDD wanting to be able  
> to separate them by having different subjects.
> 
> This is a LRDD issue not a XRD issue.
> 
> I personally think they should be separate and use the DNS name of the  
> host dns:example.com where you want a URI for the host as opposed to  
> the page.
> 
> However I am not part of the LRDD working group.
> 
> I wish you luck.
> 
> Regards
> John Bradley
> 
> On 2009-10-19, at 8:57 PM, Santosh Rajan wrote:
> 
>> Now this is exactly the point I am making, that the personal XRD and  
>> host meta are the same in the case a domain name also describes a  
>> personal resource. In the case of "thread-safe.net" your personal  
>> XRD and the host meta are the same. There is no contradiction here.  
>> It is only the context in which the resource is looked for that  
>> makes a difference.
>>
>> So if you typed in "thread-safe.net" as your OpenID, the application  
>> will simply treat the host meta as your personal XRD. On the other  
>> hand if you typed in john at thread-safe.net or thread-safe.net/john,  
>> the application will treat the XRD as host meta and look for a  
>> URITemplate with Rel="describedby" + MediaType="application/xrd_xml".
>>
>> The Rel values for your Personal Links and "general" resource Links  
>> will not be that same. There will be no overlap or contradictions  
>> here. This way we keep the whole concept clean and simple.
>>
>>
>> On Tue, Oct 20, 2009 at 4:36 AM, John Bradley <ve7jtb at ve7jtb.com>  
>> wrote:
>> Let me give an example.
>>
>> thread-safe.net is my openID so my personal XRD describing my  
>> relationship to linked resources such as an openID service provider  
>> is described there.
>>
>> I may also have other services on my site that need to have there  
>> XRD retrieved via the host-meta XRD.
>>
>> I don't want my personal XRD to have the link templates for  
>> resolving XRD on the site.
>>
>> I need something else to use for the subject of the host-meta XRD.
>>
>> Having two XRD with the same subject doesn't seem like a good idea.
>>
>> LRDD is looking for a way to indicate that the XRD applies to the  
>> DNS host as a whole rather than the URI. (For email, xmpp etc)
>>
>> You could make all http:// URL with no path "Special" but that stops  
>> people from using XRD to describe the URL itself.  At least in the  
>> openID case that would not work for many people.
>>
>> Regards
>> John Bradley
>>
>>
>>
>> On 2009-10-19, at 7:43 PM, Santosh Rajan wrote:
>>
>>> What is the difference between "describing meta data of root http  
>>> resource" and "describing meta data of the host" from a DNS point  
>>> of view? None. They are the same. It can be described by a URI.
>>> "http://example.com 
>>> ".
>>>
>>> On Tue, Oct 20, 2009 at 3:50 AM, Dirk Balfanz <balfanz at google.com>  
>>> wrote:
>>> XRD prescribes an (optional) Subject element, which is a URI. The  
>>> URI in the Subject element is the URI of the resource that is  
>>> described by this XRD.
>>>
>>> So,
>>>
>>> <Subject>http://example.com</Subject> // describes meta data of  
>>> root http resource in example.com
>>> <Subject>http://example.com/</Subject> // describes meta data of  
>>> root http resource in example.com
>>>
>>> which leaves us with the question of how to say "this document  
>>> describes meta-data data for the host example.com". The current  
>>> thinking for host-meta is to say something like
>>>
>>> <Host>example.com</Host> // describes meta-data of host example.com
>>>
>>> where the Host element is a string, not a URI. For some background,  
>>> see http://lists.oasis-open.org/archives/xri/200908/msg00127.html  
>>> and responses.
>>>
>>> Regarding civility: all-caps is not very polite. calling people  
>>> idiots is not very polite (well, I guess you merely implied it).  
>>> using lots of exclamation marks is not very polite.
>>>
>>> Cheers,
>>>
>>> Dirk.
>>>
>>> On Mon, Oct 19, 2009 at 3:00 PM, Santosh Rajan  
>>> <santrajan at gmail.com> wrote:
>>> Hi John,
>>> Let me get this strait here. I am unable to participate in the  
>>> OASIS discussions because I haven't figured the process yet. And in  
>>> any case all this has a bearing on OpenID, (it is the no 1 use case).
>>> What you are saying is
>>> 1) The host-meta will (MUST) have a <Subject> Element which will be  
>>> the domain URL of the host. There will be no <Host> element instead.
>>> 2) (This is not something you have said explicitly) . All XRD's  
>>> including host-meta "MUST" have "1" <Subject> element as an  
>>> immediate child element of the XRD Root whose value is a URI  
>>> describing the subject of the XRD.
>>>
>>>
>>> On Tue, Oct 20, 2009 at 3:04 AM, John Bradley <ve7jtb at ve7jtb.com>  
>>> wrote:
>>> Santosh,
>>>
>>> That was a thread on the use of signing elements in <Link> elements.
>>>
>>> Dirk's use of <Host> in his example XRD is not valid XRD syntax.
>>>
>>> It wasn't commented on because it was not the topic of the email  
>>> thread.
>>>
>>> If you have comments on the XRD spec.
>>>
>>> http://www.oasis-open.org/committees/download.php/34724/xrd-1.0-wd09.html
>>>
>>> You are welcome to submit them through the formal process.
>>>
>>> Regards
>>> John Bradley
>>>
>>> On 2009-10-19, at 5:51 PM, Santosh Rajan wrote:
>>>
>>>
>>> Hi John,
>>> The last time I saw an example of an XRD host-meta is here on 15th  
>>> Oct here
>>> http://lists.oasis-open.org/archives/xri/200910/msg00055.html
>>>
>>> It has a <Host> instead of <Subject>. If you are saying that it is  
>>> not part
>>> of the XRD spec and it is part of the host-meta spec, it still  
>>> doesnt change
>>> my argument. As an end-user of the the discovery mechanism the  
>>> effect is
>>> still the same for me.
>>>
>>> You say you have a hard time following me! Isn't it a case of the pot
>>> calling the kettle black? How many people are going to follow what  
>>> you have
>>> said bellow. I will only quote one sentence you have written and  
>>> ignore the
>>> rest.
>>>
>>>
>>> "The Subject of a XRD is the <Subject> of the XRD there can be 0 or 1
>>> in an XRD."
>>>
>>> That is exactly what you said. Now tell me how can there be a "0"  
>>> <Subject>
>>> for an XRD. What meaning does an XRD have with "0" <Subject>?
>>>
>>>
>>>
>>>
>>> John Bradley-9 wrote:
>>>
>>> Hi Santosh,
>>>
>>> I am having a hard time following your point.
>>>
>>> This is the current draft of the XRD spec.
>>> http://www.oasis-open.org/committees/download.php/34724/xrd-1.0-wd09.html
>>>
>>> There is no <Host>  element in the spec.
>>>
>>> The Subject of a XRD is the <Subject> of the XRD there can be 0 or 1
>>> in an XRD.
>>>
>>> HostMeta is a spec that uses the OASIS XRD spec.
>>>
>>> I know that they want to have what is essentially an abstract  
>>> Subject.
>>>
>>> ie one that is about the host and not the URI.
>>>
>>> This is a URL problem and not an XRI one.
>>>
>>> Any number of wars have been fought over how to represent non-
>>> information resources with URI.
>>>
>>> We did give the group working on host-meta as a itef spec some  
>>> options
>>> on how they might do that.
>>>
>>> Using the DNS scheme or a URI fragment are all possibilities.  I  
>>> don't
>>> know if they have come to a conclusion.   Whatever they decide  
>>> someone
>>> will be unhappy if history is anything to go by on this topic.
>>>
>>> There is a public review period for XRD coming up and a process for
>>> you to make formal submissions if you want to have input but not join
>>> the TC.
>>>
>>> John B.
>>>
>>> On 2009-10-19, at 3:27 PM, Santosh Rajan wrote:
>>>
>>>
>>> This is an Open Challenge i am sending to the webfinger, XRD forums.
>>> These
>>> guys really think I am an Idiot. "Maybe I am". "BUT I AM NOT GOING
>>> DOWN
>>> WITHOUT A FIGHT".
>>>
>>> Really, I really don't know. Let us hear the arguments they give.
>>> Maybe i am
>>> a brainless stupid, that is why i feel all of them are hollow. But
>>> let them
>>> prove I am stupid. "IF THEY CAN", IF they can, we will hand it to
>>> them, "THE
>>> IDENTITY OSCAR".
>>>
>>>
>>>
>>> Hi All,I know you guys don't like to hear from me. I have been told
>>> so much.
>>> By your moderators. That people on this forum are not "Happy" to
>>> hear from
>>> me.
>>> Like it or "NOT" you are going to hear from me. I am not sure if
>>> this post
>>> of mine will be allowed to be published. But let us see.
>>> I have so many grouses with "XRD" and today I am going to start with
>>> my
>>> first grouse. Since WebFinger by definition is going to follow XRD,
>>> don't
>>> argue with me about webfinger. Lets talk about XRD to start with me.
>>> I am throwing a challenge to all the XRD guys. Prove to me that the
>>> <Subject> of an XRD host-meta document has to be <Host> instead of
>>> <Subject>. If you "smart" guys can prove this to me, I will agree
>>> that "I am
>>> a complete Idiot". If "NOT" all of you web fingerer's and XRD's are
>>> Idiots!!!!
>>>
>>> -----
>>>
>>> Santosh Rajan
>>> http://santrajan.blogspot.com http://santrajan.blogspot.com
>>> -- 
>>> View this message in context:
>>> http://www.nabble.com/Open-Challenge-to-webfinger-and-XRD-tp25963216p25963216.html
>>> Sent from the OpenID - General mailing list archive at Nabble.com.
>>>
>>> _______________________________________________
>>> general mailing list
>>> general at lists.openid.net
>>> http://lists.openid.net/mailman/listinfo/openid-general
>>>
>>>
>>>
>>> _______________________________________________
>>> general mailing list
>>> general at lists.openid.net
>>> http://lists.openid.net/mailman/listinfo/openid-general
>>>
>>>
>>>
>>>
>>> -----
>>>
>>> Santosh Rajan
>>> http://santrajan.blogspot.com http://santrajan.blogspot.com
>>> -- 
>>> View this message in context:
>>> http://www.nabble.com/Open-Challenge-to-webfinger-and-XRD-tp25963216p25965303.html
>>>
>>> Sent from the OpenID - General mailing list archive at Nabble.com.
>>>
>>> _______________________________________________
>>> general mailing list
>>> general at lists.openid.net
>>> http://lists.openid.net/mailman/listinfo/openid-general
>>>
>>>
>>>
>>>
>>> -- 
>>> http://hi.im/santosh
>>>
>>>
>>>
>>> _______________________________________________
>>> general mailing list
>>> general at lists.openid.net
>>> http://lists.openid.net/mailman/listinfo/openid-general
>>>
>>>
>>>
>>>
>>>
>>> -- 
>>> http://hi.im/santosh
>>>
>>>
>>
>>
>>
>>
>> -- 
>> http://hi.im/santosh
>>
>>
> 
> 
>  
> _______________________________________________
> general mailing list
> general at lists.openid.net
> http://lists.openid.net/mailman/listinfo/openid-general
> 
> 

-- 
View this message in context: http://www.nabble.com/Open-Challenge-to-webfinger-and-XRD-tp25963216p26042356.html
Sent from the OpenID - General mailing list archive at Nabble.com.

More information about the general mailing list