>
>I'm sorry Fabio but it doesn't explain inconsistency between NVARCHAR, NCHAR and NTEXT() mapping to a memo field. In other words, why SQL_WVARCHAR is mapped to SQL_C_WCHAR but SQL_CHAR and and SQL_WLONGVARCHAR are mapped to SQL_C_CHAR in this case.
>
>
>When a application like VFP get ODBC data, it do this:
>

>- call SQLColAttributes  SQL_COLUMN_TYPE
>.....
>- alloc a memory buffer for the SQL_COLUMN_TYPE
>- call SQLFetch for fill the ODBC buffer
>- call SQLGetData SQL_C_CHAR <== Is the ODBC driver that copy the data !
>

>
>If VFP do not alloc a buffer with 2*N bytes,
>it cannot call SQLGetData SQL_C_WCHAR because this would generate a overrun buffer.

Fabio,

You explanation doesn't make any sense. Why would it be a problem for VFP to alocate a buffer with 2*N bytes for NCHAR and NTEXT() but not for NVARCHAR?