[openpgp] Re: Outstanding questions re Key Replacement draft

[Daniel Kahn Gillmor <[e-mail]>]

On Wed 2025-02-12 12:43:48 +0000, Andrew Gallagher wrote:

> The interim meeting did not reach a consensus on the issue of optional fallback encryption. To summarise, the basic options presented were:
>
> 0. Fallback is always possible (no change to current draft)
> 1. Fallback to each particular original is possible IFF a flag is set in the target record (granular control, wire format change)
> 2. Fallback to all originals is possible if a flag is set in the class octet (all or nothing, no wire format change)

with no formal hats on, but as an implementer and a consumer of the
spec:

I also prefer option 0.  Each layer of flags/etc is a layer to
potentially get wrong (or at least for each implementer to get subtly
different from one another, which is just as bad).  And they cause a
generating application to punt to the user some additional
very-hard-to-answer questions, which also isn't great.  And, this
particular semantic might itself at least partially duplicate semantics
we already have.

Keep it simple.

     --dkg