> The ability to correlate on eitherIn a correlated subquery there should be no need for a parameterized view>
> a) table-based udfs, or
> b) parametrized views.
since a view a subquery which can be operated on within the outer part of
the query using any required parameters. IMO table-based UDF's are too
easily used to implement procedural code where a simple subquery could do
the job. I don't really see that parameterised views would add any extra
capability. Can you give an example of the type of thing you mean?
MS intends to integrate Yukon (the next version of SQLServer) more closely> c) some sort of t-sql macro preprocessor
with .NET which may help in this regard.
Definitely useful. SQL92 permits these row-value expressions as well as> d) the ability to return multiple fields from a subquery, or
scalar expressions. I hope that feature will be included in Yukon.
Developing aggregate functions is a very complex problem which is very> e) aggregate function support for udfs
different from user-defined scalar functions. Most of the commonly sought
after aggregates (e.g. the string-, product-, mode- and median- aggregates)
have set-based solutions anyway. What sort of user-defined aggregate did you
have in mind?
Please reply only to the newsgroup