Update spec? Or is the version from 2015-08-01 the newest one?

[RokerHRO]

The spect at https://github.com/autocrypt/memoryhole/blob/master/specs/draft-memoryhole.md is incomplete, yet. Is there another, newer, more complete draft / spec somewhere?

If yes, this file should be updated (or replaced by a redirect to the newer spec). If no, the spec should be finished immediately, so it can be implemented by 3rd party software, too.

What do you think?

[dkg]

I agree that the spec needs to be updated to be completed. It is not as high a priority as the rest of the autocrypt project, though.

Of you have some time to propose improvements or fixes to the specification as specific patches, that'd be great.

[RokerHRO]

The over 2 years old draft looks more a rough collection of goals or ideas to me, and not a specification, yet.

So, sure, I can add my own ideas into that draft, but I think it would be far better when someone who knows the original plans / ideas / goals, how "Memory Hole" (I don't like that name, though) shall work, write them down.

My suggestions so far:

1. "introduction" should contain a "motivation" part, so an impatient reader knows what this document is about. (perhaps the same what is in your "abstract" paragraph above?)
2. Than I'd continue with some examples, which header lines shall be encrypted, signed, signed & encrypted etc.
3. After that the formal format description might follow
4. a section about compatibility with non-MemoryHole clients, transition paths etc.
5. suggestions how MemoryHole-enabled MUAs should create/display/check MemoryHole & legacy headers

Perhaps if there is an EBNF grammar or the like it could also moved into an appendix.

That's all from me, for now. 🙂