Sv: Implicit Access Path Sharing

1. You should rather THINK of access paths as a property of the PF than
of the LF. I.e.: "The PF has the access paths needed by it's LF's". Yes,
I know that you build the access path as the final part of the 'CRTLF',
but the mentioned thinking makes you better understand what's going on.
Direct question: Relax - all is taken care of automatically in a good
way. A shared accpth is deleted only if it is no longer in use.

2. Real world is the other way round. You already have LF1 with key A.
Then you need LF2 with KEY A & B, so you create it. I once considered
re-dooing a customers distribution system that included distributing of
*SAVF's with changed PF's and all their logical so that all LF's are
sorted by number of keys in order to establish files with highest number
of keys before files with a lower number. In that way the chance to
re-use
accpths would be higher. Worst case of access path rebuilding is now a
matter of a day or two, so avoiding a rebuild is an important matter. (A
manual solution could optimize better, but we were talking about a
one-time improvement to a utility - not a system
revision). I have however not finished this yet. Maybe there are holes in
the logic? I do NOT think that there are programs relying on seq of recs
with the same key, but maybe other holes in the logic?

3. Another topic I haven't found in this thread is DYNSLT. The general
rule for this is to use it with a low percentage of omitted records. The
limit 'low' might be moved to a little higher value if you want to CODE
for accpth sharing.


- 1 ------
> Fra: Ladutko, Bob <Bob.Ladutko@bicworld.com>
> Til: 'MIDRANGE-L@midrange.com'
> Emne: RE: Implicit Access Path Sharing
> Dato: 28. april 1999 16:38
> 
> Also, what happens if LF1 is deleted?  Does LF2 access path
automatically
> get rebuilt then, or when it's needed?

- 2 ------

> > From:       Stone, Brad V (TC) [SMTP:bvstone@taylorcorp.com]
> > Sent:       Wednesday, April 28, 1999 9:48 AM
> > To: 'MIDRANGE-L@midrange.com'
> > Subject:    RE: Implicit Access Path Sharing
> > 
> > Then the question is, if you already had LF1 with A & B as keys (and
no
> > selects, etc), why would you create LF2 with A as a key?  That's
redundant
> > to begin with and does you no good assuming there are no other
criteria
> > needed by either logical.

+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---