Timothy Meehan

Squashed commits, will merge when build is green

@hantangwangd I see that the default method for `getTableLayoutForConstraint` is performing the check that only a single layout is returned, however in theory someone could override both `getTableLayouts` and `getTableLayoutForConstraint`....

@hantangwangd my concern is someone implements both and due to a misunderstanding, they are not alerted that they have improperly coded their connector. To avoid this, I believe the engine...

> If we make getTableLayouts's default implementation return a singleton list of getTableLayoutForConstraint and still invoke getTableLayouts inside our engine and validate it's result, we might override and ignore the...

@hantangwangd I hadn't considered that. I think given that constraint, what's written here is the best we can do.

Please merge when you get a clean test run

Blocked on https://github.com/apache/iceberg/issues/9991

As discussed in #21877, I am not sure if it's most appropriate to follow the IEEE standard, or to follow the behavior of most engines. It appears that behavior of...

> I think option 3 (nan=nan and sorts largest) will be simplest to implement and maintain consistently, is reasonable behavior for users, and is aligned with many other dbs, and...

@mbasmanova thank you for the clear write up. I had one initial question. > A Prestissimo consumer would first query all (at the same time) producers for how much data...