I have a 'feature' I discovered and can't decide whether to file a bug, or a faq entry or what.
Make a TreeView which has some rows, but there is space at the end where there are no rows (i.e. the window/scrolledwindow is bigger than the TreeView). Enable drag&drop. Drag a row so that it would be dropped on the empty space at the end. Notice that there _is_ a line after the last row indicating that it will be dropped there. Drop the row, and notice that the app raises an exception and doesn't do the drop. (As an example app the one provided in http://www.daa.com.au/pipermail/pygtk/2003-November/006304.html is perfect.) Reason: the 'drag-data-received' signal's handler uses TreeView.get_dest_row_at_pos(x, y) to find out the place of the drop. In this case, however, as the drop position doesn't have rows underneath, it returns None. And TreeView.get_drag_dest_row() returns None in this case, too. (Try it out by catching the exception in the above example code and printing the values) The real suprise is that in the UI, the line appears, so one would guess the drop should work, and so should .get_dest_row_at_pos(x, y), but it doesn't. This is at least a documentation bug, as gtk.TreeView.get_drag_dest_row()'s entry in the API docs says "Returns: a 2-tuple containing the tree path and the drop position relative to the tree path" That is, it doesn't mention returning None. However, I think only fixing this as a documentiation bug isn't the best idea. Making a work-around for this isn't too dificult, as the return values in both cases should be (to match what the UI is showing with the black line): ((len(TreeView.get_model()),) ,gtk.TREE_VIEW_DROP_AFTER) (at least I think so, not tested.) so one can simply catch the exception and use these values. However, this would mean every drag&drop-handling TreeView would need this workaround, which would be really unfortunate (and not at all pythonic, and so on.) For somebody writing an App which uses dnd in a TreeView, the best solution would be if TreeView.get_dest_row_at_pos(x, y) would work in this case, too. (So this is what I would like, if it doesn't break something else.) -- Abel Daniel ps. all this with >>> gtk.gtk_version (2, 4, 4) I'm using the 2.3.94-1 .deb package from debian experimental _______________________________________________ pygtk mailing list [EMAIL PROTECTED] http://www.daa.com.au/mailman/listinfo/pygtk Read the PyGTK FAQ: http://www.async.com.br/faq/pygtk/
