Re: Request to register "identifier" relation type

[Herbert Van de Sompel <[e-mail]>]

FYI, some implementers seem to be rather OK with "identifier":

https://wiki.duraspace.org/plugins/servlet/mobile?contentId=87468598#content/view/87468598

Greetings

Herbert

On Aug 9, 2017, at 22:47, Peter Williams <[e-mail]> wrote:

>> sticking with the implemented relation name seems better than changing it to "cite" or "reference".
>  
> `identifier` would, if its use became common, almost certainly damage the usefulness of `canonical`, `self` and `bookmark` all of which *identify* the context of the link for some purpose. Most people aren't going to read the I-D. Most won't even look at the IANA link relations registry. Most will see an `identifier` link some place or skim a blog post and then start using it where ever the name seems to make sense. The usage cluster of `identifier` will be large, diffuse and will overlap the usage clusters of those other relations. The end result will be a less useful web.
> 
> The semantics described in the I-D are distinct from existing relations but the name, `identifier`, fails to evoke those distinctions. `citable` or `cite-as` would be much more likely to result in targets of these links being the resource that should be referenced when citing the context of the link. I support a relation (or relations) for the use cases identified by the I-D but not at the expense of several existing relations with proven usefulness.
> 
>> On Wed, Aug 9, 2017 at 2:00 PM, Bjartur Thorlacius <[e-mail]> wrote:
>> 
>> 
>> And sticking with the implemented relation name seems better than changing it to "cite" or "reference". 
>