[lsb-discuss] Accessibility testing in LSB

Vladimir Rubanov vrub at ispras.ru
Tue May 27 09:25:43 PDT 2008


The primary purpose of ATK tests in t2c-desktop test suite was to test the
ATK API interfaces from the LSB point of view. That is to check that libatk
library provides necessary interfaces in compliance with the ATK
documentation referenced by LSB
(http://library.gnome.org/devel/atk/stable/).

Please note that there are quite a few interfaces in ATK that are
implemented fully in libatk without regard to specific implementation
(AtkHyperlink, AtkRegistry, AtkRelation, AtkUtil, etc.). But you are right
that there are many ATK "proxy" interfaces that actually forward calls to an
implementation like GAIL/GTK. Meanwhile, LSB does not fix specific
implementation - that's why official LSB ATK test-suite should not depend on
any one and in the t2c-desktop ATK tests we used our own "dummy" ATK
implementation to be able to execute the tests in an implementation-neutral
form thus only testing libatk functions themselves. BTW, thanks to the
t2c-desktop ATK tests we have found a number of issues in the libatk library
(see
http://linuxtesting.org/results/impl_reports?action=component_bug&name=gtk-a
tk).

Meanwhile, the tests do contain checks for functional requirements and they
could be useful for testing ATK combined with a real-world implementation
like GTK/GAIL. This requires some work to adapt the tests to use particular
implementation. Unfortunately this is outside our current work focus. In the
current plan, we are going to revisit this only when we continue developing
main GTK tests in 2009.

Vladimir Rubanov
Russian Linux Verification Center
ISPRAS



> -----Original Message-----
> From: lsb-discuss-bounces at lists.linux-foundation.org [mailto:lsb-discuss-
> bounces at lists.linux-foundation.org] On Behalf Of Mark Doffman
> Sent: Monday, May 26, 2008 3:17 AM
> To: lsb-discuss at lists.linux-foundation.org
> Subject: [lsb-discuss] Accessibility testing in LSB
> 
> Hello all,
> 
> I have recently been asked to look at the Accessibility testing
> available in LSB and I have a few questions.
> 
> LSB has a different meaning of ATK to the Accessibility API of the same
> name. To avoid confusion i'll refer to the Accessibility API the long
> way as the Accessibility Toolkit.
> 
> 
> Accessibility Toolkit t2c-desktop testing
> -----------------------------------------
> 
> In the LSB repository I have found two test-suites that pertain to
> Accessibility. The first is in the t2c-desktop suite, and is a "Medium"
> quality test looking the Accessibility Toolkit.
> 
> What is the purpose of the "Medium" quality tests. The reason I ask is
> that the Accessibility Toolkit itself is just an interface standard.
> There may be multiple implementations of it, but the major
> implementation is the GAIL library for GTK.
> 
> Looking at the t2c-desktop tests I found that the implementation used is
> one crafted specifically for the tests. For example, the Object
> instantiated for the AtkComponent test a MyAtkComponent object described
> in /desktop-t2c/atk-t2c/include/AtkComponent/MyAtkComponent.h.
> 
> The first requirement for the AtkComponent interface is:
> 
> <req id="atk_component_contains.01">
>      Checks whether the specified point(with coordinates x,y) is within
> the extent of the component.
> </req>
> 
> However this requirement is not an API requirement but a functional one.
>   The implementation of this method is contained within
> MyAtkComponent.h, which is part of the test suite itself.
> 
> Is the purpose of the "Medium" quality tests simply API compatibility or
>   does it also include functional tests? Should these tests not be
> looking  at the Accessibility Toolkit API in isolation, but combined
> with an implementation? The best example of an implementation worth
> testing would be GAIL/GTK.
> 




More information about the lsb-discuss mailing list