I am milling around since quite some time on the subject of datatables. In our app we have the case that hibernates fetches a few megabytes of data, which will then be saved in the state and the hibernate-objects remain referenced -> no gc...
My current thoughts are about a datamodel that does not keep all data but just a part of the data active and requests the missing data when paging is requested... But I still have the unfinished business of the optional validation for JSF 1.1 (without facelets) if we find some developers that want to start on such a better table- component... (without committer-rights) we can share the code on a temporary base on the js-comp.sf.net project regards Alexander -----Original Message----- From: Simon Kitching [mailto:[EMAIL PROTECTED] Sent: Friday, January 13, 2006 1:11 PM To: 'MyFaces Discussion' Subject: Re: An optimised pattern for using t:dataTable - please verify. On Sat, 2006-01-14 at 00:58 +1300, Simon Kitching wrote: > Hi Yee, > > Yes, I agree with this. I think you've put it all very nicely. > > The only concern is that point (6) can be hard to get right. Oh - and also that the user will never see any changes to the database made concurrently by other users. That may be desirable, or maybe not. Of course if it is desirable, then a read before each render has to be done. However as you point out in (2), the model used for rendering really does need to be preserved until the start of the next render phase for consistency - and that saves one database read too. I would love to see t:dataTable (and maybe h:dataTable) avoid fetching the datamodel during validation when there are no input components in the table; that would make a nice project for someone wanting to get involved in MyFaces coding! Cheers, Simon
