On 10/27/11 5:42 AM, Jacek Kałucki wrote:
> I think, we should distinguish between common database tasks like querying 
> and fetching,
> which is part of querying process.
> Because of performance reasons, framework fetching process is splitted and 
> deferred in time.
> We issue query request only if filter condition changes or user request it.
> Then data are fetched as they are needed if parent context changes.
> What if I don't want to requery children on navigate, but still need every 
> node in my data
> chain get fetched after requery?
> Do I really need to move pointer and push requery button on every row?
> Solution is to use large ChildChacheInterval property value.
> BTW, big disadvantage in performance is default value of this property.
> It causes instant data requery on every record pointer move, e.g. when 
> navigating across the grid.
> By default it should be set to reasonable value like 30 sec or even more.

I agree in principle, however my testing of setting the cache interval != 0 in 
the 
past has always resulted in trouble. I never had time to dig, but I was getting 
children and/or grandchildren in the wrong context, but not all the time. I 
couldn't 
rely on it.

Perhaps recent changes have fixed it - in particular, the recent change I made 
to 
ensure that the current parent is updated even if the requery isn't going to 
happen 
could explain by prior experience.

Paul

_______________________________________________
Post Messages to: [email protected]
Subscription Maintenance: http://leafe.com/mailman/listinfo/dabo-dev
Searchable Archives: http://leafe.com/archives/search/dabo-dev
This message: http://leafe.com/archives/byMID/[email protected]

Reply via email to