Re: audit journaling source file member adds, updates, deletes

[Vern Hamberg <vhamberg@xxxxxxxxxxxxxxxxxxxxxxxxx>]

Yeah, I was both glad and surprised to find the explicit warning.

Unlike LFs, views cannot have keys. It's a little weird to us old hands at 
LFs, but is based on the SQL standard, where the data and the sorting are 
separate. The order in one of the views will be whatever comes up from 
running its SQL statement - unpredictable, unless you use an ORDER BY in 
your SQL over the view. You can see what that SQL statement is in the DSPFD 
of the view, as you may know.

Again, these views (the catalog) are here for SQL, and SQL knows noting 
about multiple members. The views filter out any multiple member files, so 
source types will not be included.

Anyway, good luck - I don't know the real risks in your case, either.

Regards

Vern

At 07:48 AM 9/15/2003 -0700, you wrote:
> Good find, Vern!  (And thanks for burning the midnight oil! <g>)
>
> Don't know what to make of the views in QSYS2.  None of them seem to be
> keyed, however using DBU shows arrival sequence, but the RRNs on the left
> side (in multi mode) are not in sequence.  Even doing a DBU search
> returned some really squirrelly results.  Thinking that DBU might have a
> problem with IBM's views for whatever reason, I tried RUNQRY
> QRYFILE((SYSTABLES)) RCDSLT(*YES) and selected TYPE = 'S' for source
> files, and the query returned no records, even though I found such records
> using DBU.
>
> Since I'm not using SQL for my plans on QADBXREF, perhaps the explicit
> statements you quoted don't necessarily apply.
>
> --- Vern Hamberg <vhamberg@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
> > Followup - SQL Reference Appendix F (V5R2) has an explicit statement
> > that
> > says not to use the QADB* files in QSYS directly. Rather to use the
> > views.
> <snip>