Several issues. First, on whether SQL (and I'm assuming we're referring to the language and interface to the DBMS) is strongly/strictly typed....it's a hard sell to say one way or the other, but certainly the compile time checking and data type constraints are there. You might do things that won't necessarily cause errors, but might cause issues downstream. On the result columns not being determined, remember you have signatures for TVFs, UDFs, TTPs, and other newer features with a certainly level of dependencies and binding. I don't think there's a compelling argument for, or against, and really the argument itself isn't a good fit for the DB platform itself.
It's on the tools themselves that consume the database where the arguments become interesting.
Ridiculous not to have typed datasets? Good grief, we are re-arguing points from years ago. And speaking of analytic packages and OLAP, most have a level of a semantic model that parallels the kind of structure you see in typed datasets. As a matter of fact, some of the BI semantic models in platforms that compete with Microsoft are even stricter than what MS provides. These features are rather helpful in larger development teams .
>
>When it comes to data analysis, there are many dedicated packages, but all of them deal with SQL sources in a dynamic (thus not strictly) typed way.
Can you name some? Again, most of the major OLAP vendors and BI vendors tend to go the other direction.
>
>You'd be right if you had to write algorithms or functions that process lots of data in a customised way (E.g. aggregation functions in SQL server written in .NET), but in the vast majority you do not need strict typing to have well performing solutions just by using SQL (E.g. SQL on local cursors), OR VFPs' xBase specialised functions.
No one is saying that strong typing leads to better performance. What I AM saying is that the common features you see in strong typing and BI semantic models have other key benefits, especially in larger teams and larger reporting applications. Not saying that strict typing in some apps isn't without pain points, but people in these environments have experienced the benefits and know how to argue against the "form over substance" misconceptions that are out there on the topic.
