OT is not patch theory. Google already had a patch-based approach (currently used in Bespin): http://code.google.com/p/google-mobwrite/, which was replaced with OT.
No diffing algorithms are used, even for producing the operations (which is probably why they're not using `contenteditable`)
Wave does allow offline interaction, and their OT algorithm is optimized for speed (for example, for the execution of possibly hundreds of buffered operations). The tradeoff is the requirement for an acknowledgement from the server after sending each operation or batch of operations.
I'm currently working on an OT implementation that doesn't require ACKs from the server in my spare time, and relies on a vector clock as opposed to a single timestamp. It has higher CPU processing power needs, which is why Wave's algorithm is also more suited to mobile devices[1]
I wasn't using diffs, my editor directly represented everything as patches, to the extent it exists at all. Much easier to do in some GUI environments than from a web browser.
Of course one of the issues is that a lot of stuff comes down to details. Working in a browser is a hell of a constraint, I wouldn't dream of writing a collaborative editor in a browser. (Not because it's impossible, but because that's an amount of stress I don't need in my life. Obviously it's possible.) It would be interesting to talk to the people involved about the issues, but I suppose that's almost always true.
(BTW, this is me interested in someone who has actually done the stuff I've only designed and curious about the exact details to a fairly deep level about the actual issues they encountered and stuff, as partially indicated in my "to the extent it exists at all" above. I am not trying to tell anyone what they should have done, I'm just chewing the fat as an interested party who stands to learn some interesting-to-me-things. I'm not in a position to tell implementors anything. As usual, tone can be hard to convey in text.)
Patch-based stuff is simply based on obsolete 30 yrs old assumptions. Why should you need a combinatorial algorithm to detect changes if you can track every user's action? etc etc
"According to an insider [10], in 6 months since the projects’s launch the possibility of working with partial histories is not implemented yet; pages are bootstrapped with their complete histories."
What I will concede is that their OT implementation is probably not perfect yet. If you read the comments on some transformation functions on their Java code, they shown concern and do extra logging in hopes to track unexpected issues they haven't solved yet.
Some unexpected issues are not solved for past 20 years.
I do not attack annotations or XML per se. The general approach "neither thing is precisely OK, so let's use both" seems questionable to me.
I believe, XML is complex enough, so why do we need XXML?
Finally, by using annotations they recognize that XML nesting is a pain in the ass, but they did not solve the problem, just sort of relieved it.
Actually, I doubt a lot, whether XML is actually carrying some useful load in the Wave. But the last one is just a feeling.
Wave does allow offline interaction, and their OT algorithm is optimized for speed (for example, for the execution of possibly hundreds of buffered operations). The tradeoff is the requirement for an acknowledgement from the server after sending each operation or batch of operations.
I'm currently working on an OT implementation that doesn't require ACKs from the server in my spare time, and relies on a vector clock as opposed to a single timestamp. It has higher CPU processing power needs, which is why Wave's algorithm is also more suited to mobile devices[1]
[1] http://portal.acm.org/citation.cfm?id=1718949 I believe one of the people involved in this paper is also involved in Wave's OT implementation paper.