[bchallenor]

The syntax for this is pretty horrible, however. And if you want to return more than one column from the subquery, you would have to duplicate the subquery definition for each column, right? Then you'd have to have faith that the optimizer can work out what you meant and reconstruct just a single subquery.

[nunwuo]

There's no faith required; the planner is guaranteed not to do that. The "normal" way is to create a composite type containing each of the columns you need, and then "unpack" it to separate columns. Horrible? Yeah, but it's possible.

[bchallenor]

Is it possible with scalar subqueries to perform anything other than a tree of correlation? With CROSS APPLY one can correlate a DAG of subqueries, e.g. a diamond where B and C depend on A, and D depends on B and C.