Re: [g-a-devel]Gnome 2.0 : accessibility for the gnome-cd application

[Bill Haneman <bill haneman sun com>]

Hi Jayaraj:

I agree with Mukund's reaction here.  Generally we don't like to NULLIFY
large parts of an API if they have meaning in a given context.  We may
nullify API that doesn't apply, but we try not to leave methods NULL
just because they are inconvenient to implement.

One point to consider is that for multi-line text a screenreader needs
to know where the text is on the screen, particularly for screen-review
mode of screenreaders.  That implies that multi-line text should always
use AtkText (rather than just the accessible name property) and needs to
implement the character extents method.

If the text has attributes that might convey any information (in other
words, attributes that aren't purely cosmetic) then AtkText is required.

In the case of gnome-cd, a possible alternative would be to expose
separate AtkObjects for each pango layout, and then you could attach a
description to each field, along with a name which carried the actual
text.

However the 'gail' library already has code that can implement most of
AtkText on behalf of a pango layout, so it may not be too much
additional work to complete the job of implementing AtkText.  For
comparison, keep in mind that GtkLabels implement AtkText.

I don't think that we currently export the headers for re-using the
pango-based AtkText implementations in gail, but perhaps we should
(rather than recommending cut-and-paste from gail).  What do you think
Padraig?  We'd have to request that this API be added, I believe. 
However if the alternative is cut-and-paste then the GNOME release team
is likely to allow it, I think.

Best regards,

Bill


Mukund wrote:
> 
> Jayaraj,
> 
> Having AtkText interface means you got to provide information on
> cursor-location, selection, getting text after position, etc. You need to
> check if these functions could be implemented for the text in the display
> area. In other words, could you get information from the application to
> support these interfaces?
> 
> I see that you would NULLIFY most of the functions in AtkText which means
> this probably is not the right interface. In other words, an AT will make a
> query on all these functions only to find that they aren't implemented,
> which is not nice to have it done that way.
> 
> Bill/Padriag?
> 
> Cheers,
> Mukund.
> 
> -----Original Message-----
> From: gnome-accessibility-devel-admin gnome org
> [mailto:gnome-accessibility-devel-admin gnome org]On Behalf Of JAYARAJ P
> R
> Sent: Monday, February 04, 2002 3:54 PM
> To: gnome-accessibility-devel gnome org;
> gnome-accessibility-list gnome org
> Subject: [g-a-devel]Gnome 2.0 : accessibility for the gnome-cd
> application
> 
> Hi,
> 
> The gnome-cd application has a custom widget called CDDisplay which forms
> the display area where the information about the time elapsed, album name,
> artist name etc appears. The parent of this CDDisplay custom widget is a
> GtkDrawingArea. It contains 4 lines of PangoLayout each of which display
> information about the time elapsed for the current track, album name and
> artist name.
> 
> As part of accessibility, i feel that this custom widget requires AtkText
> interface to be implemented. The atk_text_get_text() interface of AtkText
> would return the time elapsed, album name and artist name when the focus is
> on the CDDisplay custom widget. Most of the other interfaces in AtkText
> doesn't seem
> to make enough sense in this case and would be no-op functions.
> 
> Anyone got different ideas?
> 
> Regards,
> Jayaraj
> 
> _______________________________________________
> gnome-accessibility-list mailing list
> gnome-accessibility-list gnome org
> http://mail.gnome.org/mailman/listinfo/gnome-accessibility-list