Some Microsoft products that are based on the Jet database
engine (such as Microsoft Access) have certain requirements to be
able to update database tables. This might be true of other ODBC applications
as well. These requirements might make it necessary for you to specify
two SQL options when you define your data sources.
-
The attached table must have a
unique primary key that is not a floating-point value. You can use
the data source SQL option
Infer INTEGER from FORMAT to indicate that SAS numeric data types without fractional parts
(for example, FORMAT(
n,0),
where
n is less than 12) are
actually integer values that can be used to index the table.
-
All of the values in a row might
be used to uniquely select the row for updating. This can be a problem
in rows that contain floating-point fields (SAS numerics). Insignificant
differences in values can be caused by differences in floating-point
representation on different machines or by conversion between character
and binary formats. By specifying the data source SQL option
Fuzz Numbers at 12 places, you can cause WHERE clauses
to select values that are acceptably close, rather than requiring
exact comparisons.