Did you google?

Post #175,014 by ben_tilly 9/19/04 1:57:56 AM Reply	Did you google? I looked for "DBD::Pg serial perl" and skimmed to something that looked relevant. That took me to [link\|http://opensource.atlassian.com/projects/hibernate/browse/HB-875\|http://opensource.at...ate/browse/HB-875] which pointed at [link\|http://www.postgresql.org/docs/faqs/FAQ.html#4.15.2\|http://www.postgresq...s/FAQ.html#4.15.2] that seems to answer this question pretty well. Cheers, Ben About the use of language: it is impossible to sharpen a pencil with a blunt axe. It is equally vain to try to do it with ten blunt axes instead. -- Edsger W. Dijkstra
Post #175,015 by broomberg 9/19/04 2:19:35 AM Reply	Yup. Didn't find that though. Saw lots of people whining about coming from mysql where is has something built in for this. Hacked around it already. lock table do insert select max(id) commit - which I BELIEVE unlocks table disconnect - which I'm sure unlocks table. I'm using Sun Grid Engine to control my print spooling. I've got dozens of jobs associated with the spooling task, all of which have lots of logging to individual log files / emails to user. So I'm telling SGE to execute my logging scripts before and after each job, which should allow reporting to come off the centralized database rather than track stuff through the many log files. So the before task creates the log entry, while the after task has to update it.
Post #175,016 by cforde 9/19/04 2:40:04 AM Reply	my, that's a big hammer you have there... Yes, the 'commit' does unlock the table. It tells the database server to 'make it so'. 'rollback' undoes whatever you did, provided you haven't already done a commit. The lock ... select max... is really a brute force and ignorance (in the non-pejorative sense) approach. It will kill you on a multi-user transaction system. A data warehousing system can probably get away with it, but it is still poor form. Take a few minutes to implement the method Ben pointed to. You'll save yourself mysterious headaches later. Have fun, Carl Forde
Post #175,018 by broomberg 9/19/04 2:54:28 AM Reply	Oh, I'll modify it But the current way would NOT be a killer. The id field is indexed. So select max is not doing a table scan. Log entries instaneous from my timing point of view. The acual insert/updates might be 20-40 an hour. Each less than a second. So my hammer is not dangerous, it is just not the best way.
Post #175,020 by altmann 9/19/04 3:35:46 AM Reply	SELECT currval That's the one I've used in the past. Works like a charm. -- Chris Altmann

Welcome to IWETHEY!