RE: IBM DDS to DDL and native file structure to RDBMS tabledirection

> From: Dave Odom
> 

Quick response inline.

> He said that IBM's direction, and where their money is being
> spent, is on SQL enhancements and not to DDS when it comes to data
> structure definition.

True.  Very little is being added to DDS as opposed to DDL.  Of course,
DDL still comes up lacking in a number of areas, some of which have been
pointed out by Rob.  Another one I have had problems with is I can't
figure out how to assign an alias to a field in DDL.  Without that
capability, I can't use DDL to create a table that has short names for
native access and long names for native access.


> The same is true for data access and manipulation
> vis-a-vis SQL Data Manipulation Language (DML) vs. native READs,
WRITEs,
> CHAINS, etc.

False.

The reality is that there are two engines, and that the two engines are
both being enhanced.  The funny thing is that some SQL requests go
against the old engine, while some native requests use the new engine.
And while they may be continuing to work on SQL performance, there are a
number of things it just doesn't do as well.


> In addition, the jest was that customers should make plans
> to move both their data structure definition and application functions
> to an SQL oriented world and away from, for the most part, the native
> methodologies that have been around since S/38 days.

Only if you're an ISV, or looking to dump the iSeries as a platform.
Native I/O will always be a major component of iSeries development.  I
just got done installing a simple cilent/server architecture at a client
site: servlet invokes RPG application controller, AC invokes servers,
servers send data back to servlet via data queue, servlet invokes JSP.
Simple, clean and everything but the servlet written in RPG, no SQL.
Results?  Response time equivalent to 5250 green screen.

I'm reasonably certain that you won't get that same response time with
SQL, and especially if you wrap it in JDBC or EJBS.

Joe

Joe