RE: Externalizing input and output (record locks; unintended updates)

>The trade-off is more programmer effort to keep the logicals organized (I
>guess you could have one LF per I-O program), and you still have to deal
>with record lock considerations.  7031's get filtered with CODE/400 and it
>seems bad programming practice to change a DB data element value unless
you
>want the new value in the database.
>This topic deserves its own FAQ...
>Reeve


A couple of opinions explicit-field LF's.

When adding new fields to the file,  Most often no one goes back and
analyzes whether or not the new field should belong in the existing
explicit-field LF's.   If they do, it may be alot of "Non Value" work from
some people's  perspective(mine included.

If you want explicit control over which fields you are updating,  Try using
the Exception Time Output, with the fields explicitly defined on the Output
Statement. (This is a good practice, similar to SQL's "For Update Of"
clause.

Most shops use LF's for Index Definition more (much more) often than for
View Definition.  If that is the case(I'd say in the high 80-90% range of
practice)  Those LF's are only good for YOUR program I/O needs.   I don't
know about you but OUR end users are using THEIR data just as much as us
programmers do.  Using tools like Query/400 or whatever.   You are negating
the flexibility of the Optimizer for Their ad hoc queries in many
scenarios.  AND our(programmers)  SQL, and OPNQRYF usage.


Now about Joe's valid point of
"The wider the record length passed in to RPG, the longer it will take to
process that record."
That's true.  If you want your cake and eat it too,  more people should
have gotten excited when Han's asked "Would you like to specify a D/S in
the result field of a Chain or Read when using a EXTernally Defined File?"

I said YES!!!.   Currently that is only allowed for Program defined files.
Why is this so good?  Well,  THE WHOLE RECORD (or WFR in a New York TLA
terminology)  is brought into the D/S in ONE Move.  NOT field by field
(what Joe was pointing out).  So using that function if/when available (is
in V5R1?)  You get One Field Move I/O performance AND all fields Externally
Defined(via Ext Def D/S)


So,  Explicitly Defined Fields in LF's  ???    I stay away from them 99% of
the time.  (I also use my own Extern SQL I/O modules(


IMHO
John Carr