Interesting! as for the code generation issue:
are you aware that there is some code generation available in MyFaces? The component getters and setters, as well as the state saving, can automatically be generated... It might be interesting, though, to generate more than just the component - are you willing to mentor such a project? the second one - Is that a part of the data-model implementation which follows the specification, or is that an extended part? If we talk about specification issues, we can't do anything different there I'd say. regards, Martin On 6/3/05, Werner Punz <[EMAIL PROTECTED]> wrote: > Martin Marinschek wrote: > > Ok, I updated the proposals to include two AJAX ones, due to your > > feedback, and included a proposal for JavaScript - but note again that > > JSF can only work with disabled JavaScript if you use ServerSide state > > saving and for navigation - use commandButtons inside forms, which you > > might try to style to look like commandLinks. > > > > Please tell me if you have other proposals to make! > > > > Yes I have one, the Datamodel api currently is outright weird. > Not really weird but tailored to the smallest common denominator. > > It is tailored to forward cursors and no paging, which > a give me a row, and then increase the row count and if I am not wong > no clear boundaries, when to start and to stop a transaction. > To cope with that you usually either constantly have to open > sessions/transactions, keep the stuff open until the model is > garbage collected (which is a no go for most webapps) or keep some kind > of page cache for the deliver me the current row call (which is what I > use currently to access the data via hibernate) > > The better approach would be to have a Datamodel with a paging interface. > sort of like > > give the the number of datasets currently > give me the datasets from x to y... > > I dont see any reason why the developers went with the ... give me only > a single dataset, increase the rowcount approach, from a legacy database > standpoint which does not have paging it does not really make that much > of a difference, but the second one would make things a lot easier for > people who have to deal with that stuff. > > This approach basically would probably result in the datatable > controller code rewrite, but it should be doable depending of the > knowledge of the students. > >
