Hi Steve, Thank you for having a look at the spec. The requirement from SmartFrog was to emulate SmartFrog's concatenation of vectors, for instance, concatenating [1,2] and [3,4] as [1,2,3,4]. In CDL this is implemented as follows: <Vector1><A>1</A><A>2</A><A cdl:ref="/Vecor2/A"/></Vector1> <Vector2><A>3</A><A>4</A></Vector2> or <NewVector><A cdl:ref="/Vector3/A"/><A cdl:ref="/Vector4/A"/></NewVector> <Vector3><A>1</A><A>2</A></Vector3> <Vector4><A>3</A><A>4</A></Vector4> However, @cdl:ref cannot do the following concatenation <cdl:ref> can do <a><b>1</b><cdl:ref ref="/c" /></a> <c><d>2</d><e>3</e></c> --> <a><b>1</b><d>2</d><e>3</e></a> If we really need this type of concatenation (I would say insertion instead of concatenation), I need to bring back <cdl:ref> to the spec. We can still do that without impacting the other parts of the spec. Best Regards, Jun Steve Loughran wrote:
Dear CDDLM team and Hiro,
Please find the attached draft of CDL spec, which has been revised based on Hiro's comments. I am sorry for taking this long time for revision. The attached PowerPoint file gives highlights on this revision. In addition to incorporating Hiro's comments, I added description that clarifies the CDL's behavior when properties are duplicated (i.e.,
Jun Tatemura wrote: there
are multiple properties with the same name in the same property list). Since the Component Model uses duplicate properties, we had a discussion on the clarification several months ago. I took this opportunity of revision to incorporate that discussion. Please see the slides for details.
I'm going to take a look at seeing what the implications for this change are; ideally coding up support for it.
Overall I like the idea, but worry about
-how we support the use case of appending to an existing list -what the implications are for complex references
What is the timescale for acting on the document. Is this something to be submitted before GGF, or is it something we can talk about there?
-steve