Since we insert at the parent->child position, we need to account for

2000-11-03  Not Zed  <NotZed@HelixCode.com>

        * e-tree-model.c (e_tree_model_node_insert): Since we insert at
        the parent->child position, we need to account for expanded nodes
        above this node to properly calculate the absolute row position of
        the node.
        (e_tree_model_node_insert): If we're inserting at the end of this
        node, then we just use the position directly.
        (e_tree_model_node_remove): Completely rewritten.  Now we delete
        all nodes at once, which should be >> faster, unfortunately still
        have to signal each removal, which is >> SLOW :(  Its still about
        2-3x faster than it was (for 25K nodes).
        (child_free): Free all data/subnodes of a given path, no
        unlinking.
        (e_tree_model_node_remove): If we are removing a lot of nodes
        [>1000 or >1/4 total nodes], then use model_changed, rather then
        removing each node.  Yay.  Now its about 500x faster than it was,
        for 25K nodes.
        (etree_pre_change): Signal handler, so we can find out when we are
        in a pre-change state.
        (etree_changed): Likewise to find when we have finished.
        (e_tree_model_construct): Link to the model*changed signals so we
        know when we are in pre/changed state.
        (e_tree_model_node_insert): Only perform a row_inserted if not in
        pre_change state.  Another significant speed improvement (200-500%)
        on big trees.
        (e_tree_model_node_remove): Do not emit row_deleted (or
        model_changed), if we are in the pre_change state.
        (add_visible_descendents_to_array): Likewise for row_inserted.
        (e_tree_model_node_sort): And here too, for consistency.

svn path=/trunk/; revision=6363

0 files changed, 0 insertions, 0 deletions