Hi Ilari,

I believe your original proposal was to define a generic (polymorphic)
container that is not even dependent on HPKE.

- HSI parameter value is bstr / array / dictionary.
> - For alg==HPKE-v1-BASE, HSI parameter value is HpkeSenderInfo.


Is the following polymorphism which is limited to HPKE acceptable to you?
- "HSI parameter value can vary depending on the HPKE version and mode."
- For alg==HPKE-v1-BASE, HSI parameter value is HpkeSenderInfo.

--
Daisuke

2022年12月11日(日) 4:04 Ilari Liusvaara <[email protected]>:

> On Fri, Dec 09, 2022 at 03:01:19PM +0000, Hannes Tschofenig wrote:
> > Hi Ilari
> >
> > - Define HSI so that set of fields can change.
> >   - Parametrized by alg.
> >   - No need for registry.
> >
> > [Hannes] In your proposal HSI is an array with no extensibility
> > options. In previous discussions this was justified by saying that
> > it is fine to create a new HSI v2 when changes are needed.
> >
> > It seems that you have changed your view on the approach. Is this
> > correct?
>
> This is in line with my previous proposal. HSI and HSI v2 would have
> the same parameter identifier (which is the critical resource here).
> The recipient would use alg to decide the two apart.
>
>
> That is why the proposal specifies both that:
>
> - HSI parameter value is bstr / array / dictionary.
> - For alg==HPKE-v1-BASE, HSI parameter value is HpkeSenderInfo.
>
> ... So that when alg!=HPKE-v1-BASE, HSI parameter value can be something
> else than HpkeSenderInfo (e.g., HpkeSenderInfoV2).
>
>
> Yes, I realize that this could cause issues with some kinds of
> implementations.
>
>
>
>
> -Ilari
>
> _______________________________________________
> COSE mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/cose
>
_______________________________________________
COSE mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/cose

Reply via email to