On 13 August 2014 22:22, William Di Luigi <williamdiluigi@xxxxxxxxx> wrote: > On Tue, Aug 12, 2014 at 6:21 PM, Stefano Maggiolo <s.maggiolo@xxxxxxxxx> > wrote: > >> No, I'm *not* proposing something like: >> - creating task A; >> - cloning A to B; >> - changing field alpha of B (no change to A.alpha); >> - changing field beta of A (this change B.beta too); >> - changing field alpha of A (conflict, decide what to do). >> >> I'm proposing something like: >> - creating task A; >> - cloning A to B; >> - changing field alpha of B (no change to A.alpha); >> - changing field beta of A (no change to B.beta); >> - changing field alpha of A (no change to B.alpha); >> - you can see the differences between A and B in a "graphical diff tool" >> and move field values from A to B and vice versa with one click. >> >> This last point is what I think it would be useful for datasets too (it >> is now difficult to see what are the differences between datasets). >> > > I think this might be OK. Anyway, what do you think about Luca's proposal? > It seems similar to this. The only difference is that he proposes to clone > only the dataset when associating a task to a contest (storing dataset_id > in the TaskContest table). It seems equivalent, but cleaner. It would, > however, require the task-contest separation. > I don't like it too much, sorry - you can see that datasets are designed for another reason, because they provide the ability to change the fields that are not (too?) visible to the users; maybe now you have the same needs, but in the future might diverge and we will have problems; - in particular, there would still be the objection due to the fact that changing one task in one contest would change another contest (when you change the statement, for example); - we will have a single object doing two things.