Thanks! Your text looks good except for one inconsistency:=

"Note that integer and floating-point values are d=
istinct in this model, even if they have the same numeric value." is i=
nconsistent with "For the purposes of this specification, all number r=
epresentations for the same numeric value are equivalent." in https://cbor-wg.git=
hub.io/CBORbis/#rfc.section.3.6. This will interact with canonicalizati=
on: If 0 is equivalent to 0.0, then canonicalization has to map integral fl=
oating point values to their integer representations, to preserve the fact =
that canonicalization makes duplicate map keys easy to identify. This also =
leads to a question of whether small BigInt, BigDecimal, and BigFloat value=
s are equivalent to integer and floating values.

J=
effrey

On Wed, Sep 27, 2017 at 7:39 AM, Carsten Bormann <cabo@tzi.org> =
wrote:

Jeffrey Yasskin has proposed to wr= ite some text for CBORbis that explicitly defines CBOR's data model:

https://github.com/cbor-wg/CBORbis/issues/2

I have now written a draft for such a section, available at:

https://github.com/cbor-wg/CBORbis/= blob/master/ datamodels.md

* Do we want to have such a section?

* Is the proposed text the right one?

Gr=C3=BC=C3=9Fe, Carsten

_______________________________________________

CBOR mailing list

CBOR@ietf.org

https://www.ietf.org/mailman/listinfo/cbor