Renoir,

The firsts thing to get rid of is the form refresh() - what's the point of refreshing the grid on the second page if you don't even want to look at it (ie no match) - Just do a page1.grid.refresh() and in the activate of the second page do a page2.grid.refresh()

It may also be worth getting rid of the relationship between the 2 tables, then do an indexseek(value, .F.)
This will look for the existence of the record in the second page without moving the record pointer, if the record exists then populate the second page (with a Parameterised view or SQL statement etc)

as far as highlighting goes, the gridHighlighter classes which just affect the DynamicBackColor properties don't have much of a performance hit - Is there any other code in the grid's baseclasses to work out for example column widths manually?

There's more stuff that you can look at but that's a start :)

HTH

Will

>I've taken over a project and there are some efficiency issues that I would like some advice on. To simplify, let's say that I have 2 pages in a frame where the 1st contains a grid of a master table and the 2nd has some detail based on a child table. The current functional logic is that whenever the pointer moves in the master table (ie, grid on 1st page) it checks to see if there is a match in the child table (detail on 2nd page). If there is a match then the 2nd page becomes active, but if not then the 1st page remains active. They have seek logic and a form refresh in the AfterRowColChange method to determine the value of a logical form property. This property is then used to determine the active page. It works, but the problem is a significant delay when moving in the grid on the 1st page. I tried using eof() based on the relationship that exists, but I didn't see any change in performance. I should add that the parent class also has logic to handle highlighting the row on the
>grid. Any ideas?
>
>Regards, Renoir