Stored Procedure Compilation Question: Doing Disparate Things In Aproc

To minimize the very large number of stored procedures typically
associated with an application, I have gotten in the habit of
combining a select, insert, update, and delete all in one procedure,
and passing an argument to indicate which to use. (I use default
values for all input params to avoid having to declare them for
selects and deletes.) So I'll have just one PersonAdmin proc instead
of PersonGet, PersonInsert, PersonUpdate, and PersonDelete procs

While this is nice for housekeeping, I wonder what the compiler does
with such an architecture,and I fear the worst. The select returns a
recordset; the others don't.

Is this a bad idea?

If it is, I really wish SQL would permit some sort of user folder
structure in the proc list.