RE: Pl/sql loop assistance

2001-05-11 Thread Steve Adams
Hi Jacques, It is not reliable if there is row migration. @ Regards, @ Steve Adams @ http://www.ixora.com.au/ @ http://www.christianity.net.au/ -Original Message- Sent: Friday, 11 May 2001 3:07 To: Multiple recipients of list ORACLE-L -Original Message- From: Jared

Re: Pl/sql loop assistance

2001-05-10 Thread paquette stephane
My first rule for commit placement is the respect of the transaction (hey what do you expect from someone who likes to normalize before denormalizing ;-) ) then I looked to improve in terms of performance, modularity,... so if a transaction is done inside a loop so be it. --- Jared Still

RE: Pl/sql loop assistance

2001-05-10 Thread Jacques Kilchoer
Title: RE: Pl/sql loop assistance -Original Message- From: Jared Still [mailto:[EMAIL PROTECTED]] I'll have to disagree with not using commit within a loop. If you identify what a transaction is within your code, and write it accordingly, using commit where appropriate, you

Re: Pl/sql loop assistance

2001-05-10 Thread Jared Still
On Thursday 10 May 2001 10:06, Jacques Kilchoer wrote: This reminds me, Mr. Still, you posted an example once to the list about doing a fetch in rowid order to avoid revisiting the same block when doing a full table scan of a table, so that reading and updating every row in a table would

Re: Pl/sql loop assistance

2001-05-09 Thread paquette stephane
My observation is not on the elegancy of the code but why commit at 100 rows ? Are you updating 10 000 000 rows ? I've seen a lot of ORA-1555 because of fetch across commit. --- [EMAIL PROTECTED] a écrit : Linda, Might I suggest avoiding the elegant looping and try some inelegant

Re: Pl/sql loop assistance

2001-05-09 Thread Jared Still
Stephane, This doesn't look like it will cause ORA-1002, at least I don't see it. What's the relationship between ORA-1002 and ORA-1555? I got up rather early with a headache this morning, so maybe I'm just not thinking clearly yet. :) Jared On Wednesday 09 May 2001 05:40, paquette

Re: Pl/sql loop assistance

2001-05-09 Thread paquette stephane
Jared, They are many causes for the famous ORA-1555 Snapshot too old, one of them is fetch across commit. It is when you're commiting and fetching the same data. it is not accept in ANSI SQl but it is by Oracle. In numerous place, developpers have complained that I sized the rbs too small

Re: Pl/sql loop assistance

2001-05-09 Thread Diana_Duncan
to: [EMAIL PROTECTED] Subject: Re: Pl/sql loop assistance

Re: Pl/sql loop assistance

2001-05-09 Thread Jared Still
I'll have to disagree with not using commit within a loop. If you identify what a transaction is within your code, and write it accordingly, using commit where appropriate, you will not get ORA-1002 or ORA-1555, at least not due to your own code. I've written a number of routines to do just