Re: [evolution-patches] Re: [Evolution-hackers] HELP, question abount deleting a message in the message-list

HI NotZed,

A11Y sometimes realy need to talk to model. A11y can be image as another UI
implementation. So I deem a11y as the same level to GUI.

However, If the GUI provide some useful information, we are happy to use them. Because GUI designer do not have the responsiblity to provide all the information
for a11y layer, in most case the GUI layer information is not enough.

It seems to be this case in this bug.


Not Zed wrote:

Well do you have a proposed solution? I just said what we use in the mailer, it didn't mean it would help. I don't see why you should be using any model stuff though, since presumably for a11y you're talking about on-screen stuff, which should always go through the view.

On Fri, 2004-04-09 at 22:03 +0800, Eric Zhao wrote:

  Thank you for your replies, and sorry for my slow response, :P

  I have just browsed the codes of message-list.c(evolution) and
e-tree-memory(gal) again, the following is the source code when a
message deleted, and no row-deleted signal will be emitted. (In most
situations, etree->priv->frozen will be non-zero)

e-tree-memory.c: e_tree_memory_node_remove
       if (!etree->priv->frozen)
               e_tree_model_node_removed(E_TREE_MODEL(etree), parent,
path, old_position);
child_free(etree, path); if (path == etree->priv->root)
               etree->priv->root = NULL;
if (!etree->priv->frozen)
               e_tree_model_node_deleted(E_TREE_MODEL(etree), path);

  And when deleting message, multi messages can be seleted and deleted
at a time, so listening to the view's signal may not be enough to figure
out which rows have been deleted.


On Wed, 2004-04-07 at 10:33 -0400, JP Rosevear wrote:

On Wed, 2004-04-07 at 21:39 +0800, Not Zed wrote:
> In the mailer we listen to both the cursor changed and selection changed
> signal to find out when things change in the tree.  But that might not
> help you at all.
> > I presume the accessibility stuff should be on the view rather than the
> model right?  You're not getting any cursor changed stuff?  I think
> cursor changed used to happen when the cursor got unset too but now i
> dont think it does (why we have to listen to selection changed too - but
> that might not help you).

Does the model used by the mailer not emit a more granular row deleted
signal?  I suspect it must if speed of the delete is any indication.

JP Rosevear <jpr ximian com>
Novell, Inc.

Evolution-patches mailing list
Evolution-patches lists ximian com

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]