The whole RV v. SP argument has been overly complicated IMO. Ultimately, it is about analyzing a methodology or a technology - from this point, I will refer to them as M/T/ When I evaluate a M/T, I approach the analysis from a very selfish perspective. "What can it do for me?" "How does it make my life easier?"

Of course, the old saying of "Be careful what you ask for, you may get it..." comes to mind. On one hand, at first glance, your life gets easier. Down the road however, you find in reality, it can get more difficult and complext.

Back to analyzing M/T and what it can do for you. What I do is look at a very trivial task. I then measure the amount of work required to accomplish the same task. This is only the first step in the process. If there is not conclusive evidence one way or the other, I will then take the analysis to a more detailed arena.

The subject at hand is RV's v. SP's. Now, I need to select a trival task. The task is ordering a result set. Bascially, I am concerned with one part of a SQL Select Statement.

DB Applications built on the relational model have lookups. In a C/S model, we don't present the user with thousands of records. Rather, we ask the user to provide some search criteria and we then go an fetch records.

Both of these items serve different purposes. Each however has one thing in common. Both arise from a select statement. As time goes by, the requirements for these items becomes more refined. At first, the users might want the lists or results sorted by last name. Then, they may want them sorted by last name, then first name. Then, they may change their minds again.

This seems like a contrived example. In reality, it is not. This happens to me all of the time. The details notwithstanding, all we are talking about is changing the order by in a SQL Select Statement.

Now, lets measure the work required to effect the change for RV's v. SP's.

With a RV, I need to modify the view to reflect the new order. I then need to distribute the RV. There are at least two steps. And if there is a costly aspect to the development game, it lies in software distribution.

With a stored procuedure, I need to modify the procedure. There is nothing to distribute manually. I can make use of replication to automate the process of distibuting the change for me.

Just looking at the task itself, a SP is both easier and less costly to maintain. Anything that eliminates the need to distribute software is worth looking at. SP's provide this ability.

For me, the issue is decided. I don't need to move onto a more complicated scenario - although I could....

< JVP >