On 13 February 2017 at 19:04, John C Klensin <john-ietf@xxxxxxx> wrote: > There is one additional issue here. JMAP (or some hypothetical > xMAP) can either aspire to be a cleaner, more modern, less > LISP-like interface to (more or less, see below) the same > functionality as IMAP _or_ it can aspire to be a better idea, > with a different underlying architecture and/or reference model, > functionality, etc. > > Can't have it both ways I feel sure I'm misunderstanding something here, because clearly you can. If the underlying model of JMAP is a proper superset of that of IMAP, then it can both provide the option of different underlying architectures and/or reference models, and yet also provide a cleaner interface to the existing IMAP model. It means that some mailservices, which wish to remain perfectly compatible with IMAP, can be equally exposed over JMAP (for example, Cyrus) or exposed via a JMAP/IMAP gateway services. Yet others with a model that is different to some degree (gmail) will be exposed with higher fidelity and less awful hacks. How is this not achieving both goals here? I understand that these goals *could* be in conflict, but I think it is a false dichotomy, and one in this case that does not exist. Dave.