>Though I really want the compile to pick out those errors, it sometimes is a PITA.

Not sure if I understand the PITA expression ...

>Have you tried to write some office automation in .NET without knowing which office version the end user is using ??

I didn't but I don't know if that is a good argument to defend the issue ...

>The same applies to untyped datasets. You basically cannot be sure how a dataset looks like, until you receive it from the SQL server. Just as remote views in VFP, typed datasets are not the solution for all problems.

I'm working with 3tier design for some time now where a codegenerator generates strongly typed datasets. I really find it unsurpassed that the compiler helps me to find problems if for example I changed something in a table ...

>Don't get me wrong. compile time type checking was on my list for VFP10 as I do see that could take a lot of frustration away when making a typo. However forcing it on the compiler level is not the extreme I'm regarding as real progress.

I think it's merely a matter of taste, and we all know what is said about argueing about tasts ...