Re: QR Code data transfer protocol?

[Dirk-Willem van Gulik <dirkx@xxxxxxxxxxxxxx>]

On 17 Jul 2023, at 18:18, Carsten Bormann <cabo@xxxxxxx> wrote:

On 17. Jul 2023, at 18:09, Russ Housley <housley@xxxxxxxxxxxx> wrote:

The authors of RFC 9285 explicitly talk about QR codes.  They may be able to share more about their use cases and protocols.

Read more at https://mailarchive.ietf.org/arch/msg/cbor/M07MvOOyQlw-0P9i2GYYFd8hSbM/

Read more at https://mailarchive.ietf.org/arch/msg/cbor/gnX1E6qp0NttNjbuhephcG6BnSQ why base45 was not so bright (TL;DR: it should have been base41).

Actually - having ran the numbers - I'd challenge them - and being one of the 'poor DGC people'  I am not sure I have found out  that I was suffering much in this department. 

IMHO that particular engineering tradeoff was roughly right for Europe (with different medical models, repeat boosters and hence structures). Not in the least as a DGC contains multiple entries.

Also keeping in mind here that the need to use BaseXX-anything was that most Qr libraries where too buggy to be able to handle the Qr modes that are 8bit safe.

But given that base45 is out there, people are bound to use it for more than the (somewhat rushed) European Health Certificate.

That said -- IF (and only IF) you are willing to accept that QRs do not get overly big (i.e. barely kBytes) one can reclaim that extra half bit/10%. And if you are doing (like the Dutch) a full blown zero knowledge proof - that 10% can be meaningful enough to help.

And another tidbit found in DGC implementing -- most libraries, even on modest phones, can decode a multi kByte QR fast enough to play sequences of them as movies & get very reasonable and quite robust bandwidth that way.

Dw