Comments on: DBD::Oracle 1.22 Released

By: John Scoles

John Scoles — Tue, 02 Sep 2008 12:26:57 +0000

Sort answer: yes with a but!
Long answer: no with a maybe!

Your best bet is to give is a go and see what happens. If you run into trouble (I mean, “when”), post your question to dbi-users@perl.org You should get plenty of help there.

By: Chuck

Chuck — Thu, 21 Aug 2008 20:23:09 +0000

Hi,
I was wondering if there will be any compatibility issues when I try to install DBD::Oracle on my 64 bit machine. Perl now has a 64 bit solution and after reading throught the README.64bit.txt file for DBD::Oracle, I learned that there were problems when using a 32 bit Perl interpretor to install DBD::Oracle. If I have a 64 bit machine, 64 bit Oracle, and 64 bit Perl, will I have any issues when trying to install DBD::Oracle???

By: John Scoles

John Scoles — Thu, 21 Aug 2008 08:20:41 +0000

Yes, I have changed the code to OCIServerVersion, which will make 1 round trip to the server which in turn will keep the connection alive. This is how it is supposed to work. The original patch for for Apache::DBI was a quick hack in perl only. The 1.22 version will, I hope, use much fewer server resources, and eliminate all the problems encountered with expanding memory.

I am currently working on a new mod that will take advantage of Oracle’s own pooling system. You might want to help out with it.

Cheers!

By: Perl DBD::Oracle 1.22 Released (Christopher Jones on OPAL)

Perl DBD::Oracle 1.22 Released (Christopher Jones on OPAL) — Sat, 16 Aug 2008 01:41:12 +0000

[…] John Scoles has just released DBD::Oracle 1.22 for Perl. His release note and change list are at http://www.pythian.com/blogs/1151/dbdoracle-122-released. […]

By: DBD User

DBD User — Fri, 15 Aug 2008 12:34:10 +0000

John,

Thank you for the work you put in to this latest release. Would you mind explaining the reasoning behind the changes you made to the ping method?

I ask because one of our DBAs has singled out “select sysdate from dual” as the root cause of why our Oracle database has these random pauses intermittently. I believe the explanation was that because “Select sysdate from dual” is not able to use bind variables, it is parsed every time it executes, causing oracle shared memory to dynamically expand to accomodate “all the versions of this statement”, and in turn, causing delays for other sessions.

Would you mind commenting? I’m not sure I’ve made up my mind on this one yet and value your input. We are currently rolling v1.22 into production since the ping method has been changed, but if i’m reading the code correctly, we might just be substituting “select sysdate from dual” for a call to OCIServerVersion.