RE: Efficient Code

[G Armour <garmour400r@xxxxxxxxx>]

> I wonder if there is a way to see the indexes on a system.

Well, DBU has DBUDBR, or F14 from the main DBU screen.

On systems without DBU, you have to build your own using DSPDBR & DSPFD *ACCPTH 
&
*SELECT.

IIRC, iSeries Navigator has something related to this.  I'm too low on memory to
check it out at the moment.  

What is *really* needed, IMO, is a report you can run to identify the logicals 
that
can be "rebuilt" to take advantage of access path sharing.  Plug in a physical 
file
name, get back a listing of logicals that can be optimized like this.

If someone has already created the wheel....   <g>

GA

--- "Walden H. Leverich III" <WaldenL@xxxxxxxxxxxxxxx> wrote:
> Nope, just double checked on V5R2. If you create them in the wrong order (A,
> A+B, A+B+C) there is no sharing. 
> 
> Keep in mind too, you don't need to recreate the logical, just the index, so
> a RMVM/ADDLFM works.
> 
> Given:
> LF_A - Keyed Field1
> LF_B - Keyed Field1, Field2
> LF_C - Keyed Field1, Field2, Field3
> 
> If you create them in the wrong order (LF_A then _B then _C) LF_A owns
> index_A and LF_B owns index_B and LF_C owns index_C.
> 
> If you then you RMVM/ADDLFM on LF_B it will share the index from LF_C -
> makes sense.
> 
> Additionally, if you then delete LF_C the index ownership will revert to
> LF_B. So far so good. 
> 
> If you then recreate LF_C it will share the index in LF_B. So, when the
> index moved from LF_C to LF_B when you deleted _C it retained it's full key
> structure even though it was owned by a LF that didn't need the structure. 
> 
> I wonder if there is a way to see the indexes on a system.
> 
> Also, remember FIFO, LIFO, FCFO are all considered "keys", to get sharing
> you need to leave the ordering unspecified.
> 
> -Walden


__________________________________
Do you Yahoo!?
Yahoo! Mail SpamGuard - Read only the mail you want.
http://antispam.yahoo.com/tools