🚀 go-pugleaf

Client-side cursor fetches all data into in-memory cache and works with this
cache - because of this it can occupy many memory if query/table is large.
Client-side cursor supports filter/sort, server-side supports indeces (in MS
Jet).

//------------------------------------------
Regards,
Viatcheslav V. Vassiliev
http://www.managed-vcl.com
http://www.oledbdirect.com



"Robert Emmons" <remmons@aurigen.com> ???????/???????? ? ???????? ?????????:
news:2m65qv87t6nj09k5d6652ooultht4v2m4l@4ax.com...
> On Fri, 31 Oct 2003 20:17:07 +0300, "Viatcheslav V. Vassiliev"
> <support@oledbdirect.com> wrote:
>
> >You may use indexes defined in DBMS (inside Jet database) only if
> >CursorLocation=clUseServer and TableDirect=True.
>
> Thanks.  I had just found that CursorLocation=clUseServe causes the
> table to display in index order, but I did not know about the
> TableDirect=True requirement.
>
> I am still wondering if this is the best/only solution, since the BCB6
> help on "CursorLocation/TADOConnection" seems to indicate some
> problems.  It says:
>
> "A server-side cursor offers less flexibility, but may be more
> advantageous (or necessary) for large result sets. Using a server-side
> cursor becomes necessary when the sheer size of a result set exceeds
> the available disk space that would be needed to create the
> client-side cursor. Also, many servers only support unidirectional
> cursors. This would preclude moving the record pointer in the dataset
> backward (even one record) through the result set."
>
>