SearchFooter painting extra blank rows

In my actual project i use datasnap. However the problem can be simulated by the step below

Data

  1. Place a SqlConnection. Connectioname = FBConnection, Database =
    employee.fdb that comes with FB2.5 and other relevant settings.
  2. Place a SqlDataSet.. Commandtext = 'Select * from customer', SQLConnectio to (1) etc
  3. Place a DataSetProvider. Dataset points to (2)
  4. Place a ClientDataSet. ProviderName = (3)
    5. DataSetType  = DTNonSequenced;
    6. DataSetTypeAuto = False;
    7. Place a Datasource, DataSet point to (4)

    Grid
    1. Place DBAdvGrid on the mainform
    2. Options.goRowSelect := True;
    3. SearchFooter.Visible := True;
    4.SearchFooter.ShowHighlight := False; // pointless because can't be shown on  row selection
    5. DataSource set to the TDatasource above (7).

    When ClientDataSet.Open all the rows are populated as expected, however
    1) when scroll to the last page there's an extra blank row which is selectable.
    2) search for a value that exist in any of the columns of the row that is above the search footer or below (rows that's not visible in the 1st page).

    I have uploaded 2 images to clarify the issue further. Refer to image #1 the last row above the SearchFooter is "Cust_No 1014 Dyno Greta etc". Search for "Dyno", the cursor moves to the correct row with extra blank row, press space, backspace while editing will paint more blank rows as show in image #2.

    Image #1
    http://img230.imageshack.us/img230/8237/tmsadvgrid1.png




    Image #2
    http://img528.imageshack.us/img528/1608/tmsadvgrid2.png








Sorry (5) and (6) setting is at DBAdvGrid not the data side


These are really annoying bugs that happen almost at random. You can also look here for additional cases:
http://www.tmssoftware.com/site/forum/forum_posts.asp?TID=276&PID=1044&title=dbadvgrid-bugs-features-take-2#1044

At the moment the only thing I wait from TMS is to fix these white blanks & empty space issues since this grid is my most used component. 

Thanks Kovacevic for the pointer. I shall follow your post to unify the same problem in one post .