If any attempt is made to validate data in Image record buffers, the starting
point should be those data types already defined within Image.  This would
critically need to be controlled via some kind of configuration setting.
Lots of code in use may not obey data typing rules and would crash if type
checking were suddenly to happen (via a new Image error code?).

Following that, I'd vote for data type checking patterned after what the SQL
world does.

BIG question: how does one tell the data type from one set of bits???? It
isn't always possible to do this. One specific bit pattern can be 'valid'
data for more than one data type.

Wayne Boyer