[openpgp] Re: Call for adoption of draft-gallagher-openpgp-replacementkey

[Daniel Huigens <[e-mail]>]

Apologies for the late response.

On Tuesday, April 30th, 2024 at 20:14, Andrew Gallagher wrote:

> We could, however this would be a change in WoT semantics that people may find surprising.

I'm not 100% sure it changes the semantics as specified, since they're rather vague. But it's true that implementations would probably need to change, so you're probably right practically speaking.

> It also only works on keys with UserIDs, which we cannot assume in v6.

Fair enough. Though, for UserID-less keys, I would expect there to be some external binding/indication of which key to use for what, like WKD does for email. But, I suppose if the application does want to use a model of "implicitly or explicitly trust the first key that's presented, but require explicit indications from that key to (silently) switch to a new key", it might be useful to have an indication of that on the key.

I do think it would be better to have that indication on the new key (indeed something like a "trust equivalence" signature, perhaps), rather than on the old key, because otherwise it'll eventually be necessary to keep serving the old key purely for the benefit of this signature, while the only people who actually benefit from having this key are those who already have it locally - so we might as well only serve the signature (on the new key), to allow (eventually) removing the old key, perhaps?

Best,
Daniel