[Accessibility] Fwd: X11 libraries requested for future LSB spec

Bill Haneman Bill.Haneman at Sun.COM
Tue Jun 20 03:44:49 PDT 2006


cspi should not, in my opinion, be the ABI layer that we validate to.

As I pointed out before, the 'direct' CORBA C bindings, as generated by
an OMG-conformant IDL compiler, form an ABI in their own right. 
Similarly the IIOP, when applied to a CORBA compliant implementation of
that IDL, forms a binary protocol against which validation can be
carried out.

So the existing CORBA backend already provides us with two separate ways
in which binary validation could be carried out (i.e. two separate
binary specs on which validation tests could be built), without
involving cspi.


On Mon, 2006-06-19 at 22:07, George Kraft wrote:
> On Mon, 2006-06-19 at 17:55 +0100, Bill Haneman wrote:
> > > http://lsbbook.gforge.freestandards.org/new-abis.html
> > > 
> > > As both AT-SPI and ATs mature on Linux, then the ABIs used will need to
> > > be more strictly defined, so /usr/lib/libcspi.so ABIs and data types
> > > in /usr/include/at-spi-1.0/cspi will need to be formally specified.
> > 
> > I would argue against doing that for cspi.  I don't think we get enough
> > value from it; we should focus on at-spi's IDL interface (and the ABIs
> > that it specifies in conjunction with the OMG CORBA spec, as one of our
> > "many worlds" ABIs).
> I agree that the at-spi IDL is the best "many worlds" API.  The FSGA
> should specify the at-spi IDL API, but the LSB should follow-up with the
> complementary cspi ABI for application runtime support.  The FSG's LSB
> product standard is based on ABI runtime certification and branding.  I
> don't know how to validate an application or operating system to conform
> to or provide an IDL respectively.
> -- 
> George (gk4)
> _______________________________________________
> Accessibility mailing list
> Accessibility at lists.freestandards.org
> http://lists.freestandards.org/mailman/listinfo/accessibility

More information about the Accessibility mailing list