[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
Re: [higgins-dev] Splitting Entity UDI into absolute and relative variants
|
Based on my previous email:
* See http://wiki.eclipse.org/HOW_Update_1.1.102 <-- changes made
* See http://wiki.eclipse.org/HOWL_1.1
On 6/18/08 6:26 PM, "Paul Trevithick" <paul@xxxxxxxxxxxxxxxxx> wrote:
> With much help from Markus, Jim, and Drummond, here¹s a summary of a bunch
> of recent conversations on the list, offlist and on #higgins:
>
> - UDIs should be based on strings not URIs
>
> - We have defined some useful UDI forms
> (URIs, XRI, Linked Data, etc.) here http://parity.com/udi
>
> - But developers are free to develop their own UDI identifier forms.
> Since UDIs are fundamentally strings these developer-defined forms
> are syntactically unrestricted. E.g. the string "+1 (617) 555-1212"
> could be an Entity UDI.
>
> - For each kind of UDI there exists a UDI resolution algorithm
>
> - In the special case where the developer wishes to add a new
> developer-defined UDI form AND they wish to have these UDIs resolve
> to Entities such that they blend in with other Entities resolved by
> other UDI forms into a seamless global graph, then the resolution
> of a UDI string must be performed by attempting the multiple
> resolution algorithms in a strict order. This order will be
> defined in the UDI spec document. Developer-defined algorithms
> are applied after the algorithms/forms listed in the document.
>
> Remaining Problem
> -----------------
> - If all UDIs are strings there is no way to tell an absolute
> from a relative Entity UDI:
>
> AbsoluteEntityUDI = identifies a Context and the Entity within it
>
> RelativeEntityUDI = identifies an Entity within a given, open Context
>
> - This is a problem because some IdAS methods require relative UDIs
> (e.g. Icontext.getEntity()) whereas others require absolute UDIs.
>
> Proposed solution
> -----------------
> - We will define two kinds of Entity UDI: RelativeEntityUDI and
> AbsoluteEntityUDI. Both are string-typed.
>
> - In HOWL [1] today the higgins:entityId attribute can have values that are
> either of type string or of type EntityUDI. This will be changed
> to having a value of RelativeEntityUDI, a sub-type of xsd:string.
>
> - In HOWL we will add a definition for AbsoluteEntityUDI, as well
> as (by extension) AbsoluteAttributeUDI and RelativeAttributeUDI
>
> -Paul
>
> [1] http://wiki.eclipse.org/HOWL_1.1
>
>
> _______________________________________________
> higgins-dev mailing list
> higgins-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/higgins-dev
>