Re: SQL Question - Select * versus selecting all fields individually

There is a way to use a data structure: the SQLDA. You do not need APIs to
generate the DS, you only should have the appropriate host variables to
receive the data in.

You can use the SQLDA with (almost) any SQL statement, which I would store
in SRCPF members.

Such a utility has potential, if a green screen solution might be preferred
to tools like Sequal, Crystal Reports, and others.. It could replace all
the combinations of CDOs, QMQRYs and CLPs for reporting purposes.

But dynamic SQL has performance issues as well.

Regards,
Carel Teijgeler


*********** REPLY SEPARATOR ***********

On 6-4-2007 at 10:45 Alan Campin wrote:

For my two cents, always, always bring in only the fields you need.

Bringing in the entire table or multiple tables defeats the whole purpose
of SQL.

Also, as other have indicated, IBM does not optimize field moves. It moves

field by field. Really no reason for it to do that but that is way it does
it. The problem

is obviously important enough for IBM to provide us with data structures

on output of file I/O. SQL needs to do the same thing. If the out or input
is a data

structure, just do a single move.

This, also, opens up an idea for another utility (as if I don't have

enough to do already). Paste an SQL statement into a screen or maybe read
from a source

member and generate a data structure for that SQL Statement using the ODBC

api's. From or to line numbers in a source member or markers in the source
to >say where to start or just spin through the source member and generate
a structure for each SQL.

Does this have potential? Anybody seen a utility to do this already?