OPs to advertise support for OpenID extensions via the extension's type URI

[Andrew Arnott]

Hi folks,

Breno just pointed out to me that the UI extension's draft spec, Discovery section calls out two type URIs that should appear in an OpenID identifier's XRDS document.  But neither of these type URIs is the type URI of the extension itself.

DotNetOpenId and DotNetOpenAuth both take for granted that an extension's primary type URI (the one that appears at the value of the openid.ns.someext parameter) is expected to appear in an XRDS document if the OP supports that extension.  Maybe that wasn't a spec'd out behavior for OpenID extensions, but it seems to hold true for the OPs I tested at the time. 

While it's neat to see the UI extension include a specific Discovery section that allows OPs to declare their support for the different parts of the extension, there's no mention of declaring the extension itself.  As a result, RPs (at least the ones based on DNOI/DNOA) may not think that an OP supports the UI extension when in fact it does.  

So I'm requesting two things:

1. Can we get the UI extension DRAFT spec updated to include that the http://specs.openid.net/extensions/ui/1.0 URI be included in the XRDS document?
2. Can we standardize on the idea that an extension's type URI should be in an XRDS document if the OP supports it so that RPs can easily scan for all supported extensions? (this would be in addition to any additional type URIs the extension wants to define and advertise)

What do you all think?

--
Andrew Arnott
"I [may] not agree with what you have to say, but I'll defend to the death your right to say it." - S. G. Tallentyre

_______________________________________________
specs mailing list
specs@...
http://openid.net/mailman/listinfo/specs

[Breno de Medeiros]

I agree with Andrew on this suggestion. I don't think the UI WG proceeded differently for any particular reason, except that no such convention existed and we were not aware of side-effects previously. Regardless of interoperability issues with existing libraries, I thinking having a type URI for the extension is desirable from purely semantic standpoint (if a human were to read such document, it would be more logically organized with 'umbrella' type URIs for the extension).

On Wed, Jul 22, 2009 at 4:57 PM, Andrew Arnott <andrewarnott@...> wrote:

Hi folks,

Breno just pointed out to me that the UI extension's draft spec, Discovery section calls out two type URIs that should appear in an OpenID identifier's XRDS document.  But neither of these type URIs is the type URI of the extension itself.

DotNetOpenId and DotNetOpenAuth both take for granted that an extension's primary type URI (the one that appears at the value of the openid.ns.someext parameter) is expected to appear in an XRDS document if the OP supports that extension.  Maybe that wasn't a spec'd out behavior for OpenID extensions, but it seems to hold true for the OPs I tested at the time. 

While it's neat to see the UI extension include a specific Discovery section that allows OPs to declare their support for the different parts of the extension, there's no mention of declaring the extension itself.  As a result, RPs (at least the ones based on DNOI/DNOA) may not think that an OP supports the UI extension when in fact it does.  

So I'm requesting two things:

1. Can we get the UI extension DRAFT spec updated to include that the http://specs.openid.net/extensions/ui/1.0 URI be included in the XRDS document?
2. Can we standardize on the idea that an extension's type URI should be in an XRDS document if the OP supports it so that RPs can easily scan for all supported extensions? (this would be in addition to any additional type URIs the extension wants to define and advertise)

What do you all think?

--
Andrew Arnott
"I [may] not agree with what you have to say, but I'll defend to the death your right to say it." - S. G. Tallentyre

_______________________________________________
specs mailing list
specs@...
http://openid.net/mailman/listinfo/specs

--
--Breno

[phone] desk
[phone] (Grand Central)
MTV-41-3 : 383-A
PST (GMT-8) / PDT(GMT-7)

_______________________________________________
specs mailing list
specs@...
http://openid.net/mailman/listinfo/specs

[Allen Tom-2]

+1

Allen


Breno de Medeiros wrote:

I agree with Andrew on this suggestion. I don't think the UI WG proceeded differently for any particular reason, except that no such convention existed and we were not aware of side-effects previously. Regardless of interoperability issues with existing libraries, I thinking having a type URI for the extension is desirable from purely semantic standpoint (if a human were to read such document, it would be more logically organized with 'umbrella' type URIs for the extension).

On Wed, Jul 22, 2009 at 4:57 PM, Andrew Arnott <andrewarnott@...> wrote:

Hi folks,

Breno just pointed out to me that the UI extension's draft spec, Discovery section calls out two type URIs that should appear in an OpenID identifier's XRDS document.  But neither of these type URIs is the type URI of the extension itself.

DotNetOpenId and DotNetOpenAuth both take for granted that an extension's primary type URI (the one that appears at the value of the openid.ns.someext parameter) is expected to appear in an XRDS document if the OP supports that extension.  Maybe that wasn't a spec'd out behavior for OpenID extensions, but it seems to hold true for the OPs I tested at the time. 

While it's neat to see the UI extension include a specific Discovery section that allows OPs to declare their support for the different parts of the extension, there's no mention of declaring the extension itself.  As a result, RPs (at least the ones based on DNOI/DNOA) may not think that an OP supports the UI extension when in fact it does.  

So I'm requesting two things:

1. Can we get the UI extension DRAFT spec updated to include that the http://specs.openid.net/extensions/ui/1.0 URI be included in the XRDS document?
2. Can we standardize on the idea that an extension's type URI should be in an XRDS document if the OP supports it so that RPs can easily scan for all supported extensions? (this would be in addition to any additional type URIs the extension wants to define and advertise)

What do you all think?

--
Andrew Arnott
"I [may] not agree with what you have to say, but I'll defend to the death your right to say it." - S. G. Tallentyre

_______________________________________________
specs mailing list
specs@...
http://openid.net/mailman/listinfo/specs

--
--Breno

[phone] desk
[phone] (Grand Central)
MTV-41-3 : 383-A
PST (GMT-8) / PDT(GMT-7)

_______________________________________________
specs mailing list
specs@...
http://openid.net/mailman/listinfo/specs
  

_______________________________________________
specs mailing list
specs@...
http://openid.net/mailman/listinfo/specs

[John Bradley-8]

+1 I think that advertising the extension itself is a good practice.

A RP may prefer OPs that support the extension over ones that don't.

That is the case for PAPE now as an example.

With XRD most of that will be described in the OPs XRD rather than the  
users, but the same principal should apply.

John B.
On 22-Jul-09, at 12:00 PM, specs-request@... wrote:

> From: Breno de Medeiros <breno@...>
> Subject: Re: OPs to advertise support for OpenID extensions via the
> extension's type URI
> To: Andrew Arnott <andrewarnott@...>
> Cc: specs@...
> Message-ID:
> <29fb00360907221019t10a0393aydbae458ba8c662ba@...>
> Content-Type: multipart/alternative;
> boundary=00151750e13a821afc046f4e91df
>
> --00151750e13a821afc046f4e91df
> Content-Type: text/plain; charset=ISO-8859-1
> Content-Transfer-Encoding: 7bit
>
> I agree with Andrew on this suggestion. I don't think the UI WG  
> proceeded
> differently for any particular reason, except that no such convention
> existed and we were not aware of side-effects previously. Regardless  
> of
> interoperability issues with existing libraries, I thinking having a  
> type
> URI for the extension is desirable from purely semantic standpoint  
> (if a
> human were to read such document, it would be more logically  
> organized with
> 'umbrella' type URIs for the extension).

_______________________________________________
specs mailing list
specs@...
http://openid.net/mailman/listinfo/specs

[David Recordon]

Sounds good to me!

On Jul 22, 2009, at 5:23 PM, John Bradley wrote:

> +1 I think that advertising the extension itself is a good practice.
>
> A RP may prefer OPs that support the extension over ones that don't.
>
> That is the case for PAPE now as an example.
>
> With XRD most of that will be described in the OPs XRD rather than  
> the users, but the same principal should apply.
>
> John B.
> On 22-Jul-09, at 12:00 PM, specs-request@... wrote:
>
>> From: Breno de Medeiros <breno@...>
>> Subject: Re: OPs to advertise support for OpenID extensions via the
>> extension's type URI
>> To: Andrew Arnott <andrewarnott@...>
>> Cc: specs@...
>> Message-ID:
>> <29fb00360907221019t10a0393aydbae458ba8c662ba@...>
>> Content-Type: multipart/alternative;
>> boundary=00151750e13a821afc046f4e91df
>>
>> --00151750e13a821afc046f4e91df
>> Content-Type: text/plain; charset=ISO-8859-1
>> Content-Transfer-Encoding: 7bit
>>
>> I agree with Andrew on this suggestion. I don't think the UI WG  
>> proceeded
>> differently for any particular reason, except that no such convention
>> existed and we were not aware of side-effects previously.  
>> Regardless of
>> interoperability issues with existing libraries, I thinking having  
>> a type
>> URI for the extension is desirable from purely semantic standpoint  
>> (if a
>> human were to read such document, it would be more logically  
>> organized with
>> 'umbrella' type URIs for the extension).
>
> _______________________________________________
> specs mailing list
> specs@...
> http://openid.net/mailman/listinfo/specs

_______________________________________________
specs mailing list
specs@...
http://openid.net/mailman/listinfo/specs