HP3000-L Archives

January 2003, Week 5

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Gavin Scott <[log in to unmask]>
Reply To:
Gavin Scott <[log in to unmask]>
Date:
Thu, 30 Jan 2003 17:11:33 -0800
Content-Type:
text/plain
Parts/Attachments:
text/plain (64 lines)
Ok, my thoughts on this are:

1) Roy is the only one on the list who clearly understands what's going on.

2) Extended Sort fields have no issues *whatsoever* with CIUPDATE.

If you DBUPDATE a path with an associated sort item (a "critical item"),
then I presume that Image correctly honors the extended sort items when
re-inserting the entry in the path.  To do otherwise would require a
separate special insert-into-sorted-path routine to have been written to
explicitly *prevent* it from honoring the extended sort items, which I
assume was not done.  If I'm wrong on this, then DBUPDATE should be fixed to
do the insert respecting extended sort keys.

DBUPDATE has always messed up extended sorts if you alter one of the items
comprising the extended sort key, and again this has nothing to CIUPDATE at
all because extended sort items are not "critical items".  Using extended
sort fields has always required one to avoid using DBUPDATE on parts of the
extended sort key.

3) If the default for CIUPDATE is changed to "ON" then I presume that this
will only be a change to DBSCHEMA, which means that only newly created
databases will be affected.  To do otherwise would require removing the
CIUPDATE mode configuration parameter which would *prevent* someone from
turning it off.

A change that affects only new databases is a relative waste of time
because:

a) To a first approximation, all Image databases that will ever be created
have already been created.

b) It introduces an obscure inconsistency between any newly created
databases and all existing databases.

4) The current default of "ALLOWED" means that any program that wants
CIUPDATE can have it via a DBCONTROL Call with no other action on the part
of the user.

To gain a benefit from "ON" requires that:

a) People who could benefit from "ON" are too stupid to have turned it on
for themselves.

b) People were smart enough to try DBUPDATE before falling back to
DDBDELETE/DBPUT and have no bugs in their DBUPDATE logic that might cause
problems if the DBUPDATE starts working.

5) If DBUPDATE is changed to always respect extended sort keys, this is a
major change to Image which will have unnecessary negative performance
implications for all users of Image who do not care about extended sort
items.

6) Everything is working fine today.

Therefore, in my own opinion, there is no pressing need to change the
behavior of DBUPDATE, or to change the current default setting of "ALLOWED"
for the CIPDATE mode.

G.

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *

ATOM RSS1 RSS2