Bruce Collins wrote:
> Roy wrote:
>> The only way round this would be if the program had code to check if
>> the extended area was changed, and always forced a DBDELETE/DBPUT if
>> so. But if this was the case, it would never fall into a CIUPDATE in
>> these circumstances....

> I think we are talking about two different problems.

> In the case where the critical item itself is updated it sounds as it
> the DBUPDATE is adjusting the position in the chain without
> respecting the extended sort items, whereas, if the record is
> modified with a DBDELETE/DBPUT the new record will be inserted in the
> proper location in the sort chain.

I don't think this is the case. Remember that a record can be in up to sixteen
chains, and a critical item need not necessarily be critical in all the chains
through the record. I conjecture that if the DBUPDATE needs to adjusts the
position in one or more of these chains, it *will* respect the extended sort
items. But for the chains it  does not need to adjust, it won't respect these
items, because there it will behave like a non-critical DBUPDATE.

But it doesn't really matter; either way, we agree that it is not guaranteed
to respect extended sort items. But that's not because of CIUPDATE; it's
because of DBUPDATE - as you go on to observe, precisely (and concisely!)
nailing the issue I was trying to raise:

> The problem with updating a non-critical item which is part of an
> extended sort is a problem with DBUPDATE whether or not CIUPDATE is
> turned on. One solution to this problem would be if IMAGE was able to
> recognize these items as critical in addition to the actual sort item
> and then drop into CIUPDATE mode (if it is turned on), or disallow
> the update of the (extended) critical item if CIUPDATE is off.

> Perhaps now we are talking about a third enhancement request.

It's certainly the precise issue. But I think this is actually the true shape
of what the second request should be (the first being Wirt's default of
CIUPDATE to ON).

--
Roy Brown
Posting with the OEnemy, tamed by OE-QuoteFix 1.18.3
http://jump.to/oe-quotefix

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