>>>I have a class built that communicates with a hardware device. This device may be connect via serial, ethernet or ftp. The properties of the class set all the necessary parameters of the device... tcpip/serial/ftp, address, port, bps, parity, etc. The class is instantiated, properties are set (based on settings stored in a device table) and the communication takes place. This works great. But occasionally after the class is instantiated and the properties are set, the device needs to be changed to a different one. This requires access to the device table. I don't like the idea of tying the class to a table to query it. I have don't have any ideas on how to keep this class abstract yet still query the data in my device table. Any ideas?
>>
>>Make the name and path to the table a property of the class, so you can change it at runtime?
>>
>>Tamar
>
>Yes but then I have to hard code field names and queries into the class. That doesn't seem like a good idea does it?

That table is your metadata. It is code, just packaged as a table. It's under your control (or if it isn't, it should be).

And you can package the queries as scripts in memo fields of that table, so whatever is specific to one device should be in that device's record.