Re: Uniquely identifying a record in SQL without a unique key?

[Charles Wilt <charles.wilt@xxxxxxxxx>]

As others have mentioned, if your talking about DB2 for i, there's rrn()...

Note however, that contrary to logic...using WHERE RRN(tbl) = 1
causes a FULL TABLE SCAN, not random I/O!

You'll find some similar functions in other some other DBs, Oracle for
instance has the rowid() function.

But MS SQL Server doesn't officially offer anything comparable, but it
looks like there's at least 1 unsupported method...
http://stackoverflow.com/questions/909155/equivalent-of-oracles-rowid-in-sql-server
But again, using the above solution forces a full table scan.

Also as mentioned, when using a cursor to read a results set, you can
UPDATE WHERE CURRENT OF...looks like JDBC 2.0+ offers "updatable resultset"
JDBC 1.0 used the actual WHERE CURRENT OF syntax

Basically, it boils down to the idea that physical location of the row
isn't supposed to matter in an RDB and that having tables without a unique
key is bad.

I've used tools in the past that simply don't offer update functionality
for tables without a unique key, including IBM's own CPYF  MBROPT(*UPDADD)
:D

Personally, that's the route I'd take.

Charles





On Fri, Dec 28, 2012 at 1:54 PM, James H. H. Lampert <
jamesl@xxxxxxxxxxxxxxxxx> wrote:

> I'm beginning to study Scott Klement's JDBCR4 API, as a means of
> extending our QuestView product to access external databases (and it
> would also allow us to expand our very limited support of data types
> that are normally SQL-only), and something suddenly struck me:
>
> Suppose I'm looking at a record, and I decide to make a change to it.
> And the file isn't uniquely keyed.
>
> Now, with native RLA, we can easily determine the RRN of the record
> we're looking at, and acquire an update lock on that record by its RRN.
> After all, RRNs only change if a RGZPFM is done, and that requires an
> exclusive lock on the member.
>
> But now, suppose I have to use SQL, JDBC, and the JDBCR4 API. I'm no
> expert on SQL (SQL for Dummies is still my primary reference on the
> subject), but I don't recall ever reading anything anywhere about a way
> to get a record's RRN, or retrieve a record by its RRN, from SQL.
>
> I must be missing *something* here, but what? How do I guarantee that an
> update will go to the right record, when I initially read it without a
> lock, and apparently can't grab a lock on it by its RRN?
>
> --
> JHHL
> --
> This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
> To post a message email: MIDRANGE-L@xxxxxxxxxxxx
> To subscribe, unsubscribe, or change list options,
> visit: http://lists.midrange.com/mailman/listinfo/midrange-l
> or email: MIDRANGE-L-request@xxxxxxxxxxxx
> Before posting, please take a moment to review the archives
> at http://archive.midrange.com/midrange-l.