Re: [core] Comments on draft-bormann-core-links-json-01

[Carsten Bormann <[e-mail]>]

On Jul 29, 2012, at 18:37, Zach Shelby <[e-mail]> wrote:

> Do we want this JSON format to be a generic representation of RFC5988 Web Links, or is this something really only specific to the CoRE Link Format use of Web Linking? I think deciding on that scope would help close some of the open questions in the draft. Personally I find the later purpose really nice, but looking at the bigger picture this should probably be as generic as possible.

I think the guiding principle should (as in most protocol design be):

Generality is great if it costs nothing (or very, very little).
When generality starts to make things complex or lose some of the focused goodness of a proposal and it's not quite clear the the important use cases require this complexity, you are almost always better off doing the less general thing.

http://en.wikipedia.org/wiki/You_ain't_gonna_need_it

So, my questions would be:
-- Do we know about any complexity that would come from making this a general RFC5988 representation?
   (Such complexity may include the need for interfacing with other groups, which can slow us down unless we know how to do it efficiently.)
-- Do we lose something in the process?

If we can be reasonably sure both are not the case, we should go for the whole thing.

Grüße, Carsten