[urn] Re: URN Namespace Registration for SAID (Self-Addres sing Identifiers)

Hi Dale

The hash/digest algorithm is encoded in the SAID, at least partially, for cryptographic agility. Applications may use different digest functions over a long period of time. Data with different digests may also merge. Self-encoding allows the overall system to evolve gracefully. Another example is PQC.

In some sense, we could argue the same thing for canonical serialization too (and that’s part of the motivation for CESR).  However, the application space is likely to be much more complex and diverse of course. There are already many formats and numerous data schemas around. The current design basically says if you add a SAID field, you are always having the hash algorithm information already embedded. We ensure that the two come together. The SAID field can also be separated from the container data, and later reunited, and so on.

Regards
Wenjing


From: Dale R. Worley <[email protected]>
Date: Thursday, January 29, 2026 at 6:48 PM
To: Wenjing Chu <[email protected]>
Cc: [email protected] <[email protected]>, [email protected] <[email protected]>, [email protected] <[email protected]>, [email protected] <[email protected]>, [email protected] <[email protected]>
Subject: Re: [urn] Re: URN Namespace Registration for SAID (Self-Addressing Identifiers)

Wenjing Chu <[email protected]> writes:
> The new application is here:
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Ftrustoverip%2Fkswg-said-urn-charter%2Fblob%2Fmain%2Furn_application.txt&data=05%7C02%7Cwchu%40futurewei.com%7Cd4eea51fe5364c25b55b08de5faa07c5%7C0fee8ff2a3b240189c753a1d5591fedc%7C1%7C0%7C639053381053436026%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=Zwx2HuWvfSlYR3MHWWoN%2F9nNCRUdMQWr5CP4c87vGmE%3D&reserved=0<https://github.com/trustoverip/kswg-said-urn-charter/blob/main/urn_application.txt>

I don't see any reason to not register this namespace.

However the definition of it seems peculiar to me.  The registration
(and the definition of SAIDs generall) is very careful to have the SAID
contain a code specifying the hash that is used.  As the registration
says "each primitive includes a prepended type code identifying both the
cryptographic algorithm and value length, enabling stream parsing
without external delimiters".  But the process for verifying a SAID
requires that the application knows the canonical serialization for that
class of data objects and the canonical manner in which the SAID
placeholder is inderted into the object before it is hashed.  Given the
amount of information the application must share a-priori with other
applications which generate/verify a SAID, it's not clear why the hash
algorithm is not part of that information.

Dale

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