RE: Simple Display file problem

["Alan Campin" <Alan.Campin@xxxxxxxxxxxxxxxx>]

<snip>
Actually, I'd suggest learning the appropriate places to use the
different
database tools that the iSeries gives you.  SQL is appropriate for
set-based
logic, typically queries, while single record fetches and especially
record-at-a-time updates are much better written using native I/O.

</snip>

 

I might agree with you Joe except for one thing and that is database
independence. Because IBM does not give us an easy way to bring in just
what we need from a table, SQL is a better solution for just about
anything because I can. 

 

Every time we bring the entire table into a program with record I/O, we
are hard coding the physical view of the data into the program and every
program that uses that table has to be recompiled every time we add a
new field. 

 

We can easily add new fields in DB2 now except for breaking every
program using the file in record I/O. 

 

The other issue, for me, with using record I/O is increased complexity
in the program. 

 

Almost invariably, when I look at I/O in a program, I find relationships
and instead of creating one SQL statement that brings in several tables
at the same time, we end up chain to one and then another and then
another and looking at values. All of this can be done in one SQL
statement and the database can do the work that end up being hard coded
into the program.  

 

By continuing to use the record I/O model, we continue to train
ourselves to think in terms of the single file and not in terms of
database.

 

With free-format SQL in V5R4, things finally get easier. If IBM would
get the SQL runtime to the ILE call model instead of the OPM call model,
we would really be cooking.

 

Well, this ought to heat up the wires.