Ok, I know anyone has a custom solution for this very problem but I think we should find a general one, why reinventing the wheel? We've a couple of great paging taglib so I think we need to interface to them instead of recode everything.
1- The taglibs need the data stored in the session (which is not that bad, perhaps it doesn't scale exceptionally) 2- The data should be a collection of beans (usually) 3- the user really doesn't need more than 15/30 records at once so I think we need to code a "virtual collection", this object simulates millions of records but only has 30/40 records cached, when the taglib request a record and the collection gets a miss (a not cached data) it should retrieve new data from the db (using the application db pool), same thing when the taglib request a sort. The code should be general enough to interface to various db and pools. What do you think? -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
