Funny how you can work with data a long time and still run into something new. I’ve been working on an upgrade for a third party application with a few moving pieces – you know how that goes. One of the pieces is a batch processor and we struggled to get it running, and finally it took a call to support to figure out why. The application requires a DSN (which is really a server and database name, essentially a portion of a connection string) and we had done that. We created as a system DSN, then as a user DSN, and we just weren’t getting a connection, or even a failure.  It turns out that the application requires a 32 bit DSN, and that bit of magic requires running a different ODBC applet, this one in the syswow64 folder.

I was astounded. Maybe there is some really good technical reason for this – if so, I’d like to hear it. It seems like adding a ‘check this box if you need 32 bit compatibility’ in the main ODBC applet would be have been easy enough, and a local more easy to understand. Beyond that though, the distinction escapes me – it’s still just a server name and  database name and maybe a few miscellaneous settings. I didn’t dig to see if stored in a different registry key, I’m assuming it is – again,for reasons I’m not clear on.

I’m writing about because maybe someone can enlighten me,but also to perhaps save you the hour I spent getting to the bottom of it!