Hi Jim,
Thank you for this update. I have a couple of questions about the IANA
registries in the algorithms draft.
* Why did you set the "Capabilities" of COSE Key Types registry to "kty, x "?
In particular, I do not see why having "kty" in this column would be helpful at
all. The kty value for each line in the "Capabilities" column should be the
same as the value in the "Name" column, i.e. "1, crv" for Key Type 1, "2, crv"
for Key Type 2, "3" for Key Type 3 etc. I suspect you did add it to match the
"Parameters" column of section 9.2 of OSCORE groupcomm, but if that's the case,
having only the second element of the current column would be enough. We could
then define in OSCORE groupcomm the format as "kty (taken from "Name"), X taken
from Capabilities for that kty".
* I would expect this document to define exactly how the fields in Capabilities
are formatted, and what values they take. It is enough to have a sentence
saying "crv is defined in section TBD, and can take the following values: TBD
from this registry: TBD". Also, the reference column of this registry should be
updated to also point to this document.
* I have a hard time understanding the update to COSE Algorithms registry. You
have added a column called "Capabilities", and populated every raw for all the
algorithms with the string "kty"? How does that help? Maybe you wanted to point
to the COSE Key Type to use with that algorithm? That would mean adding a "Key
Type" column (rather than a "Capabilities"), populated with the value from the
"Name" or "Value" column of COSE Key Types registry, for each algorithm. Then
we would extract the capabilities for that algorithm by looking at the
Capabilities for the right key type. Another option would be to duplicate the
information from "COSE Key Types" Capabilities into the Capabilities column of
COSE Algorithm that you define, but I don't like duplicating things. If we
cannot assume that "Capabilities" are the same for the algorithm and for the
key type, I would add the "Capablities" column in the COSE Algorithm as well.
Even then, I do think a "Key Type" column with the "Name" or "Value" from COSE
Key Types would be useful. That is the way to link: "You use this type of key
with this algorithm".
Example of what I think this should look like:
COSE Key Types registry:
Name | Value | Capabilities | Ref
1 | OKP | crv | This doc
2 | EC2 | crv | This doc
Crv field for Value OKP takes the values from Table 22 of RFC8152 for Key Type
OKP
Crv field for Value EC2 takes the values from Table 22 of RFC8152 for Key Type
EC2
etc
COSE Algorithms registry:
Add a "Key Type" parameter, that takes for value for example:
Name | ... | Key Type
EdDSA | ... | OKP (or 1)
etc
Hope this makes sense,
Francesca
On 11/11/2019, 23:53, "COSE on behalf of Jim Schaad" <[email protected] on
behalf of [email protected]> wrote:
I have finally gotten to the point of sending this message.
I updated three drafts before the submission deadline
* The X.509 and hash algorithm drafts were updated to address WGLC comments
* The Algorithms draft was updated with an attempt to get a more
generalized version of what was needed by the group communication algorithm
details.
I need to get feedback on the new capabilities defintions to see if I have
missed anything significant and to see if this will substitute for what was
desired by the group communication authors.
Jim
_______________________________________________
COSE mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/cose
_______________________________________________
COSE mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/cose
