[MlMt] JMAP support

[Randall Gellens]

On 11 Jan 2019, at 12:10, Sam Hathaway wrote:

> On 11 Jan 2019, at 14:58, Randall Gellens wrote:
>
>> An IMAP-JMAP proxy just moves the complexity of dealing with the 
>> myriad of IMAP servers from core MailMate to an embedded proxy.  I 
>> don't see it providing that much help, while it would undoubtedly 
>> introduce its own set of problems.
>
> If coordinated right, it would centralize the thankless work of 
> kludging around nonstandard, broken IMAP server implementations. 
> Imagine if, instead of each MUA author having to develop, test, and 
> maintain dozens of ugly hacks, they were collaborating to improve a 
> single JMAP proxy codebase that all could use.

That could be a great thing.  Or it could be just one more variant for a 
client to support, potentially multiple variants if the proxy behaves 
differently depending on which IMAP server it is facing.  Done well, a 
universal proxy could help.  But then, done well, IMAP is fine.

>> As I said earlier, while JMAP might be very cool, it doesn’t help 
>> the core problem of widely variant IMAP server behavior; instead, it 
>> just introduces yet more variants.
>
> It does move us towards solving the problem of IMAP being kinda trash 
> for resource-constrained clients with slow network connections.

As I mentioned before, IMAP was originally designed for extremely slow 
dial-up lines that were prone to sudden disconnection.  Many of the 
features of IMAP are specifically for the problems of communication in 
that environment.

> I think it’s telling that many of the major email providers (Google, 
> Microsoft, and FastMail, at least) felt the need to create their own 
> proprietary mail access protocols for use by their mobile apps. What 
> this says to me is that IMAP is not fit-for-purpose when it comes to 
> smartphone apps.

Maybe, or maybe many of the big providers didn't want to spend the time 
to figure out how to do IMAP well, and/or saw competitive advantages of 
having their own protocol.  There's a long-standing problem of players 
big and small doing poor jobs of implementing standards.  In some cases, 
it was clear that developers read the RFC examples but not the normative 
text.

--Randall