Accessibility API status w/e Dec 8

[Bill Haneman <bill haneman sun com>]

Hi Folks:

Here is the API status for accessibility modules this week. 

ATK:

Made AtkRole into an extensible enum.  Exported this along with other
ATK enums via atk-enum-types.h (generated file), so that language
bindings can use them (python bindings for instance).

There were no other API changes this week.

However we have identified an issue with signal emissions, at the moment
there is no way to identify the AtkObject from which an interface-based
signal originates, only the interface instance.  This is problematic for
event notifications, it appears that an API revision to ATK will be
required in order to implement our event notification in at-spi.  The
simplest solution would be to move these signals from the interfaces to
AtkObject, though this might seem to lose some specificity.  In any case
there will never be an instance of an ATK interface that does not have a
corresponding AtkObject instance, so this may actually better reflect
our intention.

AT-SPI:

libspi : removed accessibleeventlistener.c in favor of eventlistener.c
(just a rename, including a rename of the interface names).  Started
removing libspi headers from installed headers list, probably we should
not install any of these headers, the public interface to libspi should
be via CORBA, thus just the IDL.

idl : 

Added getLayer and getMDIZOrder to Accessibility:Component, to reflect
previous additions to ATK.

removed dubious attribute 'hash_id' from Accessibility:EventListener
Changed Accessibility:StateSet::compare () to return a difference
StateSet rather than use an out param.
Changed Accessibility:Image::getImageExtents () to return a
Accessibility:BoundingBox,
and changed Accessibility:Component::getExtents () to do the same. 
Introduced structure Accessibility:BoundingBox.

We left the getPosition, getSize methods as-is, but may choose to
deprecate them soon as redundant with getExtents.

cspi: 

Changed use of 'boolean' to 'SPIBoolean'.  Some restructuring of the
code, but not a lot of impact on the C API as used.  

Best regards,

Bill