Re: [dispatch] Proposal for Per Resource Events Protocol

[Michael Toomim <[e-mail]>]

Dale:
> My memory is that few or none of the SIP event types that support
> partial updates consider the state to be divided into a sequence.
> Usually, it's a set of key/value pairs, where the key is a string that
> identifies the component of the overall state (e.g., the identifier of a
> particular call) and the value is some complex structure.  So if you're
> considering partial updates, start with having a way to specify
> modifying the values for individual keys.  Indeed, you may need little
> more than that in practice.

Yes, in OT and CRDT research, Key/Value pairs are the easy things to 
synchronize.

It gets about 100x harder as soon as you have a text document with 
multiple editors. Then you need to synchronize a sequence of characters 
(or DOM nodes, or whatever), and you need sequence synchronization. 
Sequence sync algorithms are 100x harder than key/value sync.

But it's pretty awesome when you make it all work! ;-)