Given that we occasionally look at how other SDOs use CBOR, COSE, etc.:

This week here in Europe, everybody talks about (Covid-19) “vaccination certs”.

European Digital Health Certificates (as used in the European Digital Green 
Certificates “DGC” for Covid-19) use CBOR and COSE.

This development had to happen quickly, so you it was run in the open; you find 
lots of information in github.
It is entertaining and educational to read the discussions on the github repos, 
e.g.,

»It's very easy to decode HCERT. I coded our HCERT verifier in 4 hours without 
knowing anything about CBOR and COSE (I have used LD-Profs in the past, but not 
COSE-based certs).« [1]

It is also interesting that one of the problems they had was with QR-Code 
integration.  E.g., the base45 ASCII-funneled encoding they are using [2] 
contains percent characters, which unsurprisingly get mangled by some 
smartphone QR-Code readers (who try to percent-decode them as URIs).  (That 
would have been easily avoidable at zero additional cost by using a 
base41-style encoding instead and using a less risky charset subset.)

Also, they are using zlib (deflate) to “compress” the COSE, except that it 
doesn’t [3].

We may not pay much attention to these integration issues in our IETF WGs, but 
they are really important to make the whole package work.

Maybe we can find someone to talk about practical aspects of DGC and related 
efforts at one of the next CBOR interims...

Grüße, Carsten


(Thanks to Emmanuel Baccelli for alerting me to this…)

[1]: 
https://github.com/ehn-dcc-development/hcert-spec/issues/64#issuecomment-830692153

[2]: https://datatracker.ietf.org/doc/html/draft-faltstrom-base45

[3]: https://github.com/eu-digital-green-certificates/dgc-testdata/issues/284

_______________________________________________
COSE mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/cose

Reply via email to