Adrian Holovaty wrote: > Here's the current status: Robert is going to finish up manipulator > tweaks by the end of the week, when he leaves for a week-long > vacation.
So I'm off tomorrow. The current status afaik is: Core fields removal. core=true and related things now give a validation error. Edit inline is now done via explicit add and delete commands in multiple requests. This actually works (at least locally) a lot better than I thought it might. It should be fairly easy to make this work with js : it should just hook the add and delete buttons, and add/delete the form controls client side. An issue remaining here is that html2python does not get called for added inline objects right now, so it tries to save using the unconverted post data. I have a hack fix for this. Fixing it nicely requires messing about with formfields a bit - I think I will leave this for when I get back. The other issue is the FileField and ManyToManyField things I sent a message about before. I will attempt to do these tonight. > After that, Jacob is going to Ajaxify the admin edit-inline > objects. After that, we'll evaluate descriptor fields and implement > that if we decide to go ahead with it. Finally, we'll start > encouraging people to test out the branch, and, if all goes well, > it'll be in good shape for merging. > > There are other features in our to-do list, such as the UnitOfWork > concept (essentially transactions), but that's not > backwards-incompatible. Can anybody think of other > backwards-incompatible changes that we'd want to roll into > magic-removal before trunk? Well, I guess the subtyping thing. Although the backwards incompatible bit ( breaking the copy subtyping ) has already happened.
