Am 02.11.26 um 18:10 schrieb Richard Clayton:
What Ed25519 has shown us is that almost no-one will implement ANY new algorithm for DKIM1 because there is absolutely no use case for doing so and engineering effort will be applied to DKIM2 instead. Some people have not even got the memo about SHA-1 or short RSA keys.
Some open-source products have implemented Ed25519. Anyone who wants to use it can do so, but is bound to the respective software. Additionally, based on experience, the number of DMARC reports containing DKIM errors (due to a lacking Ed25519 compatibility) is decreasing.
The issue of incentives and engineering effort is precisely why we picked a PQ algorithm for DKIM2 at this stage of its development and put it into the spec with MUST instructions. We can change which PQ algorithm we choose up until Auth48 -- after that to all practical intents and purposes we will be stuck with our choice until DKIM3 turns up in the mid 2050s.
I think the MUST is the correct approach for DKIM2, it could also be the case for DKIM1. Bastian -- Bastian Buck - Geschäftsführer BWeb-SSL Services UG & Co. KG - 88512 Mengen - www.bweb-ssl.de AG Ulm HRB 736203 - Geschäftsführende Komplementärin: BT computers & more UG (haftbeschr.) E-Mail: [email protected] - Webseite: www.bastian-buck.de _______________________________________________ Ietf-dkim mailing list -- [email protected] To unsubscribe send an email to [email protected]
