Authors,
I added a comment to the C509 ticket #222 [2] but I will restate on the mailing 
list for a more direct audience.

When looking at the possibilities for a `C509CertData` bytes container I ran 
into what seems like a lack of specification for the input data to the hash 
which would be used to generate `COSE_CertHash` values for a "c5t" header. This 
makes it unclear whether the input to the hash would be a full encoded 
`C509Certificate` (including array head) or the sequence `~C509Certificate` 
(excluding array head, which is what is proposed for the TLS certificate type) 
or something else entirely.

In either case, my opinion is that the `C509CertData` be defined so that the 
byte string is exactly the same as whatever the input to the hash function is. 
This makes life easy for implementations to be able to construct a C509 
certificate hash without needing to decode or inspect the byte string at all.

[2] https://github.com/cose-wg/CBOR-certificates/issues/222

> -----Original Message-----
> From: Sipos, Brian J. <[email protected]>
> Sent: Monday, March 17, 2025 4:03 AM
> To: Carsten Bormann <[email protected]>; Anders Rundgren
> <[email protected]>
> Cc: Göran Selander <[email protected]>;
> [email protected]
> Subject: Alert-Verify-Sender: [COSE] Re: [EXT] I-D Action: 
> draft-ietf-cose-cbor-
> encoded-cert-13.txt
> 
> Anders,
> Sure, I think either would be helpful to separate the "transport of C509" from
> "decode/inspect the C509". A similar discussion [1] was had in the DTN WG
> about bstr-embedding of other data for similar reasons, but hasn't yet 
> reached a
> conclusion. In the case of [1] the container array size was fixed so provided 
> no
> information. For C509 a definite array size at least can inform a decoder 
> about
> how many extensions are present before decoding them, but in practice I don't
> know how valuable that is for implementations to have.
> 
> [1]
> https://mailarchive.ietf.org/arch/msg/dtn/mmfQGd8K2W2PaH02RJDPIMMl52g/
> 
> 
> > -----Original Message-----
> > From: Carsten Bormann <[email protected]>
> > Sent: Friday, March 14, 2025 1:25 PM
> > To: Anders Rundgren <[email protected]>
> > Cc: Sipos, Brian J. <[email protected]>; Göran Selander
> > <[email protected]>; [email protected]
> > Subject: Re: [COSE] [EXT] I-D Action:
> > draft-ietf-cose-cbor-encoded-cert-13.txt
> >
> > APL external email warning: Verify sender [email protected] before clicking
> > links or attachments
> >
> > On 2025-03-14, at 18:11, Anders Rundgren
> > <[email protected]>
> > wrote:
> > >
> > > bstr(~C509Certificate)
> >
> > I don’t know which notation this is supposed to be, but unwrapping a
> > C509Certificate in CDDL as in ~C509Certificate creates a group.
> > Is the intention maybe
> >
> > bstr .cborseq C509Certificate
> >
> > ?
> >
> > (Which saves one byte from
> >
> > bstr .cbor C509Certificate
> >
> > )
> >
> > Grüße, Carsten

Attachment: smime.p7s
Description: S/MIME cryptographic signature

_______________________________________________
COSE mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to