Now that we have scalar udf inling in SQL Server 2019 our scalar udf:s can run fast. Before SQL Server 2019, I always rewrote slow scalar udf:s to table valued udf:s.
I would say that writing and using a table valued function instead of a scalar one is only slightly more work. Very slightly. So the practical use for the feature "scalar udf inlining" seems quite limited. OK, I'll save some keystrokes but isn't there more to gain?
Are there many or any cases where a scalar udf would be more useful than a table valued function?