Re: [COSE] How COSE items are identified

I think it’s premature to assume one structure is used “most of the time” as opposed to other structures and to optimize COSE for that case. Though if we take a cue from JOSE, it’s asymmetric signing, then MAC. Those might be reversed as I don’t have concrete numbers on this. Public keys are probably next. Encryption is far behind in my experience with JOSE.

My point is that I don’t think we can pick any one structure as “most common” yet as we’re extrapolating from exactly zero real-world examples.

 — Justin (as non-chair)

> On Nov 16, 2015, at 1:12 AM, Jim Schaad <ietf@augustcellars.com> wrote:
> 
> We seem to have a potential problem for how COSE items are identified.
> 
> Initially, we used an integer inside of the structure to identify it.  That
> is how the current ACE documents are written. This uses a single byte.
> 
> At the request of Carsten, we moved from doing this to using a tag at the
> front of the structure.  Carsten believes that this can be done with a tag
> that is in the two byte range when encoded (tag, length=1, tag value).  Part
> of the reason that I agreed was that this was not really being used at the
> time by the ACE documents.  
> 
> If one is almost always doing encrypted, then one could encode that
> structure w/o the tag and do the other three structures with the tag.  This
> would probably be ok as it would mostly use the smallest tagging (none) when
> needed the most and use larger tagging (2 bytes) for the rest of the
> structures.
> 
> Alternatively, we could modify the document to request the assignment of
> tags in the one byte range for the most common structures and two bytes for
> the less common structures (and have arguments about which is which).
> 
> Alternatively, we could revert back to putting an integer tag inside of the
> array structure.
> 
> Comments on this are requested.
> 
> Jim
> 
> 
> _______________________________________________
> COSE mailing list
> COSE@ietf.org
> https://www.ietf.org/mailman/listinfo/cose