[rfc-i] The alternateURI element in v3

Julian Reschke julian.reschke at gmx.de
Thu May 8 08:23:31 PDT 2014


On 2014-05-08 17:18, Paul Hoffman wrote:
> On May 8, 2014, at 7:57 AM, Julian Reschke <julian.reschke at gmx.de> wrote:
>
>> On 2014-05-08 16:50, Paul Hoffman wrote:
>>> On May 7, 2014, at 11:38 PM, Julian Reschke <julian.reschke at gmx.de> wrote:
>>>
>>>>> - Added <alternateURI> to allow reference to things such as DOIs.
>>>>
>>>> I don't believe a "type" attribute is needed.
>>>
>>> Assume someone wants to find the DOI of a document. Also assume that the URI for DOIs changes in the future. Having the "type" attribute allows them to know definitively what the RFC Editor thought was the DOI without them having to know all historical URI patterns.
>>
>> I thought we were discussing DOI URIs as defined in RFC 4452?
>
> We were discussing DOIs, and then someone (maybe you?) said that instead of just a DOI attribute in <rfc>, we should have a new element for alternate URIs of any type.

Yes.

>> In which case detecting a DOI URI is just a case-insensitive scheme match.
>
> True. But, if that's the only alternate URI we need, it could be an attribute called "DOI".

I can imagine all kinds of alternate URIs :-)

>>> So, maybe it is not "needed", but it is useful for people who want to find a specific type of alternate URI.
>>
>> I'm afraid that we are adding a new typing system without clear requirements, and ignoring existing work.
>
> And I'm afraid you are stretching work that is for MIME documents past its breaking point.
>
>> If we need typing on links, why not use the type system established by RFC 5988?
>
>
> In what possible way? We don't have MIME headers here: we have an XML element. The name of the element is exactly in line with the "alternate" type from 5988. You seem to be objecting to the "type" attribute that meant to be used for searching; 5988 doesn't deal with that, I believe.
> ...

RFC 5988 defines linking. It hasn't to do anything with MIME. It *does* 
define an HTTP header field, but that's only one way to represent a link.

The HTML equivalent is the <link> element, which has a "rel" attribute. 
One trivial approach would be just to adopt that as a top-level 
document, and to define those relation types we need (such as "doi").

See also 
<http://greenbytes.de/tech/webdav/rfc2629xslt/rfc2629xslt.html#ext.element.link>.

Best regards, Julian


More information about the rfc-interest mailing list