[Accessibility] FSGA Teleconference Agenda, Wednesday 28 June
Bill.Haneman at Sun.COM
Wed Jun 28 10:32:38 PDT 2006
I am sure we'll talk more about it tonight, but I personally don't think
we have only the two options you suggest. I would support adding the
AT-SPI CORBA ABI to LSB as part of a conformance approach that
explicitly allows for additional conforming ABIs in the future.
I think we can do better than just giving up on conformance testing for
the time being. A standard with no conformance test doesn't seem very
meaningful in the LSB/FSG context.
Unless I am misunderstanding you, your suggestion would ensconce cspi
and perhaps later the pyton bindings, "QSpi", etc. as additional
requirements in LSB. That seems to me to be more restrictive to distros
and to our efforts than a roadmap that allows us to migrate backends
later while not tying us to a C api that we don't particularly like,
just because it's convenient.
On Wed, 2006-06-28 at 18:25, Olaf Jan Schmidt wrote:
> I won't be able to attend today, so let me just provide a comment to the
> agenda by email.
> [ Bill Haneman ]
> > * LSB requires validation, so we must relate the
> > 'normative IDL' notion to an actual ABI
> Is there a requirement to add AT-SPI to the LSB?
> My suggestion is that we publish AT-SPI as a separate standard with different
> validation rules from the LSB. The aim would not be to define an ABI, but to
> ensure that the functionality is there.
> For the LSB, I see basically two options:
> a) We postpone all discussion of standardising an ABI until we have an ABI
> that we are all comfortable with. The functional spec will still allow us to
> check wether AT-SPI support is shipped in a distribution. Since the number of
> AT-SPI variants will be limited to two at the most, it does not seem
> necessary to rush defining a frozen ABI.
> b) We add cspi to LSB. At the same time we allow distributions to use
> additionally include other interfaces to AT-SPI (and use them in the
> assistive technologies they ship). Once other CORBA-independent AT-SPI
> bindings emerge (e.g. a Python ABI or a QSpi ABI) these are added to the LSB
> as well. Assistive technologies then have the option of using an LSB-approved
> ABI if they wish to, but we still have the option to develop better ABIs.
> Since Peter and Bill are most familiar with cspi, they should be able to say
> best which of the two approaches makes more sense.
> Olaf Jan Schmidt, KDE Accessibility co-maintainer, open standards
> accessibility networker, Protestant theology student and webmaster of
> http://accessibility.kde.org/ and http://www.amen-online.de/
> Accessibility mailing list
> Accessibility at lists.freestandards.org
More information about the Accessibility