>>Experimenting with pulling dbfs into SQL database, via select * from openrowset(...), I've found that ALL of the memo fields have been migrated with an extra chr(0) in the end.
>>
>>Googling a subject where your keywords are "null", "null terminated string", "vfpoledb" and such gives an incredible amount of fun, seeing how many completely unrelated things pop up, and an even larger amount of frustration.
>>
>>So... the question: did anyone encounter this before? If not, is there a good place with information on how VFPOLEDB can be configured? Or, at least, what can go into the provider string to change this behavior?
>
>
>If memo or char field contain CHR(0), the field type on SQL must be binary and in query you must convert memo/char to
image(varbinary(max))/varbinary.

This is the opposite problem. The fields are just text; when imported from VFP into SQL via openrowset(), they end up having a chr(0) added at the end. Something along the chain thinks this is supposed to be a null terminated string.

Problem temporarily solved by using ODBC driver instead, but I'd still like to know what's causing this and whether it can be avoided.