Jon Gjengset

I don't know that the goal is 100% compatibility, in part because this is a research prototype, and many MySQL features simply aren't particularly interesting from a research perspective. Operators...

We should probably document what bigger things are currently missing though @ms705

The biggest challenge we have is that the join algorithm has to be incremental. Since it's a dataflow system, this means only certain subsets of the more general join problem...

That is correct — there is no such documentation at the moment. Since this is a research prototype, we sort of just support what the implementation supports, and it's not...

Ah, yes, you're right, there is also no documentation of the high-level details about queries. In general though, it is just standard SQL; that's why we didn't think to do...

@ms705 I agree with @mitar that we should documented how Noria's SQL syntax translates to Noria's concepts. For example, things like the `QUERY` keyword and `CREATE VIEW` operations would be...

@HenkPoley Correct :)

Sorry, when I say "full materialization", I mean full materialization for the live query, not full materialization as in "all results of all queries". Where this distinction comes up is...

Right, so that was my proposal — a mechanism would have to be added for the application to register interest in particular values for `z`. Once that's in place, _then_...

@mitar It's true that Noria could provide that feature too, though that still has the same issue: Noria won't know when a key that hasn't been materialized changes, and so...