[quoted text, click to view]

¤ Hello Paul,
¤
¤ Thnx for your quick answer.
¤
¤ I knew that solution, but I would like to avoid the use of the schema.ini
¤ file since the procedure will be used by different types of (typed) datasets.
¤ Isn't there another way to tell the OleDBAdapter to first check all rows
¤ (and thus the column types) before creating the datatable?

Unfortunately there isn't much choice. In the absence of a data type specification the Text driver
will guess at the data type depending upon the predominant values in the column. In your example the
predominant values (in the rows scanned) would appear to be integer instead of a floating point data
value.

You could try changing the MaxScanRows entry in the Registry to a value of zero (which is supposed
to cause a scan of the complete file) but I can't recall whether this was actually reliable.

HKEY_Local_Machine\Software\Microsoft\Jet\4.0\Engines\Text\MaxScanRows

MaxScanRows can also be used in a schema.ini file.


Paul
~~~~