|
>Thanks for the test. I hope I didn't come across too testy (so to speak) >when I threw down that gauntlet. Must be this cold. Not at all! Like I said, I was happy to do it... this sort of thing is always an education... [The question is this: If I issue and SQL request against a logical] [and the logical is built with select/omit criteria, will the records] [returned be filtered by the s/o criteria?] I ran some tests: >> >>OK. Here's the set-up: >>Physical file MASTER keyed by PEXC, PNUM, PSUF. Record count 13,390. >>Logical file BILLPRT keyed by ZIP, BTEXC, BTNUM, BTSUF, PEXC, PNUM, PSUF >> SELECT TYPE NE 'C' >> >>If I run this statement: select count(*) from master >>I get this answer: 13,390 >> >>If I run this statement: select count(*) from billprt >>I get this answer: 13,322 >> >>If I run this statement: select count(*) from billprt where type = 'C' >>I get this answer: 0 >> >>In all these interactive statements, I get the following message in debug: >>Arrival sequence access was used for file xxxxxx >> >>I ran this one: select * from billprt order by pexc,pnum,psuf >>Got this count: 13,322 (I scrolled to the end of the list) >>Got this message: Access path built for file BILLPRT. >> >>On my machine at V4R1, SQL used the logical I specified, and it obeyed >>the select criteria in use. If there's another set of circumstances you'd >>like me to test with, I'd be very happy to do it: I would have been >>shocked to the core if I did a SELECT from the logical and got all the >>records, under any circumstances! > >Looks like it had to create a new access path, since the ORDER BY you used >is deep into the logical's key. > >What results do you get if you specify the WHERE clause with the physical, >setting the same criteria as the S/O above? > >select * from master where type <> 'C' order by pexc,pnum,psuf Just what'd you'd expect: 13,322 records (same as the S/O logical) The optimiser timed out, and built an access path even though the physical file if already keyed the same as the ORDER BY. Buck Calabro Commsoft +--- | This is the Midrange System Mailing List! | To submit a new message, send your mail to "MIDRANGE-L@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 +---
As an Amazon Associate we earn from qualifying purchases.
This mailing list archive is Copyright 1997-2025 by midrange.com and David Gibbs as a compilation work. Use of the archive is restricted to research of a business or technical nature. Any other uses are prohibited. Full details are available on our policy page. If you have questions about this, please contact [javascript protected email address].
Operating expenses for this site are earned using the Amazon Associate program and Google Adsense.