> > > I am confused on which DBD to choose, though. Should I use > DBD::ODBC > > or DBD::Oracle? I can set up ODBC DSNs on each computer > without much > > problem so both are equally convenient. But is there any > other reason > > or criteria I can use to make this choice, perhaps based on > > performance or stability? > > The way I've heard it is: always always use the native Oracle > libraries as much as possible, avoid using ODBC for Oracle. > So stick with dbi:Oracle:*. >
Ok -- here's my explanation as the author of DBD::ODBC. I use DBD::Oracle whenever possible, it's faster (typically) and works on my multiple platforms (Linux, Solaris and Windows). I can do DBD::ODBC from all these platforms, but using a bridge. I can do DBD::Oracle natively from all these platforms. If you are only using Oracle, then I'd use DBD::Oracle. However, there are times where you *know* you need to support SQLServer or Access or whatever and your script needs to handle date/time or call stored procedures. In those cases, you can use the ODBC syntax for date/time and function calling which *may* make your porting easier. Also, if you are going to rely upon DBD::ODBC for other database access, you should test with DBD::ODBC for Oracle too... In either case, if you find something that works with DBD::Oracle and not DBD::ODBC, I'd want to know. It may be something esoteric, but, I'd like to fix it or document it. Regards, Jeff
