[Kde-accessibility] Re: glib dependancy in KDE3.x

Olaf Jan Schmidt ojschmidt@kde.org
Thu, 6 Mar 2003 18:39:51 +0100


=2D----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

[Philippe Fremy]
> I remind or inform you that glib will also be a dependancy for the
> accessibility module, planned to be implemented as a plugin. So we have
> actually  three projects which are going to depend on glib.

Just to put things right: The KDE Accessibility Project is NOT planning to=
=20
introduce a glib dependency into KDE.

We are discussing about a bridge between AT-SPI and Qt that will depend on=
=20
both ATK and Qt. This bridge will have no dependency of glib, only of=20
ATK, and it need not be part of KDE itself - especially since it is=20
planned as a ATK-Qt bridge rather than a KDE glib dependency.

The second thing I wish you to keep in mind is that we have not started to=
=20
do any actual coding. We are still in the phase of discussing possible=20
strategies, and we are currently concentrating on implementing smaller=20
projects.
Apart from the bridge, AT-SPI support requires heavy changes to both Qt=20
and KDE that will be discussed on kde-core-devel when the time is right.

We will not start coding the Qt-ATK bridge before the Qt and KDE side=20
works at least basically, and we will not start changing KDE before we=20
have come to an agreement with TrollTech and kde-core-devel, and we can=20
only start those discussions when we have have a clearer picture how=20
AT-SPI, ATK and QAccessible can exactly work together, which is not easy=20
since all of them are very poorly documented.

It does not make much sense to decide KDE core issues upon a planned=20
accessibility interoperability bridge that is not even been coded yet.

> I think that for the projects where it makes sense, glib dependency
> should be accepted. It is a pity to waste developer's time developing
> something that exists elsewhere in better and maintained quality.
>
> What are the projects for which it makes sense ? Sofar, we've seen two
> cases:
> - a KDE project wants to talk to a gnome project. This is the case for
> arts/gstreamer and kde-accessibility/at-spi.
> - a project is developed within the scope of KDE but can not depend on
> Qt because the scope of the project is wider than KDE. That's the case
> for wv2.
>
> In both cases, the glib dependency makes sense.

I agree that a glib dependency would make some interoperability issues=20
easier, but at least for the KDEAP there is currently no need for a=20
direct dependency. Everything can be solved via bridges.

Accessibility interoperability will require more then just basic KDE and=20
Gnome installations. Special software like the AT-SPI registry or certain=20
device drivers are also needed, so asking for the installation of a=20
bridge is exceptable for people who need AT-SPI.

If kde-core-devel decides that using glib is OK, then we might change our=20
plans, but we currently are only talking about an ATK dependency for the=20
bridge. And, to say it again, we have not started coding anything, so=20
everything I say is just ideas and possibilities, and might be changed=20
later in favour of a better solution.

Olaf.

=2D --=20
Olaf Jan Schmidt, KDE Accessibility Project
KDEAP co-maintainer, maintainer of http://accessibility.kde.org

=2D----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iEYEARECAAYFAj5nh+gACgkQoLYC8AehV8fO8gCfVWnTLbezMDtgUH5A88b3FUbF
c4AAoMW5hBqvYiIzFe+L0Rr6jNylP2/H
=3DLi+S
=2D----END PGP SIGNATURE-----