[lsb-discuss] LSB: finishing off the Qt4 question

[Olaf Schmidt]

Hi!

[ Wichmann, Mats D, Fr., 20. Jul. 2007 ]
> - Qt4 integration with other technologies, particularly
> the evolving accessibility framework (atspi)

In LSB 3.1, we made atk mandatory with a number of in-process tests, but 
without any AT-SPI-based out-of-process tests.

The equivalent approach would be to make Qt 4 mandatory with the in-process 
tests that already exist. We can make a D-Bus-based out-of-process test for 
QAccessible mandatory later (Trolltech told me that they are already working 
on it internally) just as we did with atk/gtk.


The consensus in the OpenAccessibility workgroup was that we cannot make 
AT-SPI support mandatory for LSB libraries until its inherent problems have 
been fixed. For example, there is no maintainer for one of its dependencies 
(ORBit2), and some parts of it are too GNOME-specific to be used as a general 
accessibility framework. Without substantive changes, it does not meet the 
requirements for LSB standardisation.

AT-SPI can be temporarily used for better atk conformance testing until a 
common solution has evolved, but I must object to any standardisation of 
AT-SPI itself.

We do not want the LSB to make any ABI stability guarantees for AT-SPI. It 
needs to be clear that future conformance tests of the LSB will likely use an 
incompatible version AT-SPI (or an alternative to AT-SPI). If an alternative 
framework evolves and if distributions start to migrate to the new framework, 
then the LSB must allow them to use different conformance tests based on the 
new framework. This warning needs to be either in the spec itself or in a 
place whether it can be easily found and referenced in future discussion on 
AT-SPI and accessibility.

Olaf