Hello, I have created a sample application which demo's similiar to the myfaces pagedSortTable demo:
http://www.devosc.com/pub/myfaces-cars.zip This demo has a new type of DataModel class called a PagedDataModel, which will work with a ListDataModel, and presumably an ArrayDataModel, it might work with the ResultDataModel but will not work (atleast not straight away) with the ResultSetDataModel because of its getRowCount method implementation. This demo I believe is the most practical for real usage given the wiki documentation. [1] http://wiki.apache.org/myfaces/Working_With_DataTable_And_PreserveDataModel As far as I can now tell there should be caveat that, that the discussion is only in regard to having the whole dataset in memory, given: "as the data will be retrieved directly from the previously serialized data" [2] http://wiki.apache.org/myfaces/WorkingWithLargeTables The method setWrappedData is disabled making it not possible to update the dataModel when sorting the dataTable and trying to work around it causes duplicate queries to the database. The following are observations made of the demo about the number of db queries queries performed (potential in this case) in a requests lifecyle given the respective settings, saving state method, bean scope and preserveDataModel. [2] client, request, false [2] client, request, true [2] server, request, false [2] server, request, true [1] client, session, false [1] client, session, true [1] server, session, false [1] server, session, true The number of calls to create the actual data model is denoted within the square brackets. Each test was redeployed and the browser session cookies removed. So from the above, the only setting that ensures that the database will not be queried more than once when either scrolling or sorting a dataTable is if the managed bean is in the session scope and it would also seem that the preserveDataModel attribute is of no consequence. I did try to use the 1.1.2-SNAPSHOT jars but a javascript error occurs related to the dataScroller. I would be interested in using/having the latest jars that allow the demo to work. What I would like to understand is why there are two queries being performed when the bean is in the request scope ?
