Hi there! just some thoughts ...
I also think that it is right about time to include dojo into tomahawk. Since a lot of sandbox components are already using dojo an probably a lot more will come this should be the right way to go. Sooner or later those components will move from the sandbox to tomahawk. For the updates I just want to point out that we should probably maintain a list of dojo-using components somewhere to go through and make sure affected components are tested after a update. Unfortunately I don't know a way of writing unit-tests for HTML rendering components which I guess would be the ideal thing for that. regards Ernst On 8/19/06, Werner Punz <[EMAIL PROTECTED]> wrote:
Hi I just wanted to start an open discussion about the dojo related roadmap. My plan is following for now. Now that we have 0.3.1 in we are going to fix our pending components. Then we should aim for a dojo inclusion into tomahawk core for 1.1.6. (if we move the codebase out of our sourcebase into maven is still open, as it seems weblets wont make it until 1.1.6... for now) then we can start to integrate dojo to reduce our own tomahawk dhtml codebase and to get rid of some long outstanding issues. The main problem for now I see is, that sun and others have started to use dojo as well, and we might get a version clash in the long run once we start to mix tomahawk with the creator components in the long run. Dojo itself has not solved that issue, I am not sure how to avoid those problems except for trying to be always on the latest stable version and to hope that the others follow this route. Anyway in the long run we also will have to be in sync with the resource loading mechanisms, while dojo is pretty good in its resource loading we really should avoid double includes and different loading namespacing for the same things. Due to the fact that we route everything on the java side through tools classes this change is possible afterwards. But as I said, moving dojo towards Tomahawk is pretty important to reduce our own codebase (for instance the date picker dhtml codebase probably can be dropped entirely as well as the popup code, which has been a constant problem for years.) and to fix some long outstanding issues. The interfaces seem stable enough to promote them out of the sandbox once we are done with the upgrade. But the downside is, have in mind, that the dojo upgrades will come once every few months, while the interfaces are kept pretty stable, once we have moved Tom towards dojo we will have the problem of having to do some work after the upgrade on many of the tom components, but I think the problems are way less than the benefits of having a lot of less dhtml codebase on our side.
