Dave McAlpin wrote:
>Thanks for summarizing this Gabe. I'll just make comments on two of the
>examples
>
>xri://naming.authority/local/part/(+version/1.2)
>
>It seems wrong to me that 1.2 is an immediate child of +version in that it
>pollutes a namespace we intend to use for version types. An example like
>this might be better.
>
>xri://naming.authority/local/part/(+version/numeric/2)
>
>
>
I would prefer if we also allowed for version types themselves to be
dereferencable. That way, we can accomodate an open versioning
framework, rather than attempting to define all possible version types.
This is the way Liberty handles the Attribute-portion of the identity
framework (ID-WSF: SIS).
so a valid version string may look like:
xri://naming.authority/local/part/[+version/urn:version-authority:dottedalphanum/1.2a]
or even:
xri://naming.authority/local/part/[+version/urn:xri:version-authority:dottedalphanum/1.2a]
etc... of course, a version-spec referenced as
xri://version.name.authority/spec/namespace may be difficult for parsers.
--- peterd
