Re: Debian specific accessibility issues

[Bill Haneman <Bill Haneman Sun COM>]

Jason:

You are correct that GNOME code has a review process whereby patches are 
accepted into cvs.  Of course several GNOME accessibility folks have cvs 
commit access already - the basic structure is that each GNOME 'module' 
in cvs has one (or preferably more) "maintainer" who is responsible for 
the module.  I am currently the maintainer of most GNOME cvs 
accessibility modules, but not gnopernicus or orca.

We already have patches for the bugs I mentioned below, stored in the 
gnome bugzilla bug tracking system.  However, it's up to the maintainer 
of the relevant module to review the patches and either approve them for 
committal by the submitter, or to apply them directly to cvs.  If the 
maintainer does not review these patches in a timely manner, or objects 
to them, then they do not get applied.  A similar process applies to 
Mozilla, and as you probably know, the code review process for Mozilla 
substantially slows the application of the accessibility patches to 
cvs.  However, code review serves an important purpose as you well know.

Any distro maintainer is welcome to review the GNOME Accessibility 
Buglist which Calum Benson maintains at 
http://gnome.org/~calum/access-bugs.html, and may wish to apply the 
patches which are available in bugzilla for many of these bugs, and 
which are already being applied to Sun's JDS distribution.  As a rule I 
don't think any other distros are doing so yet.

Your other suggestion, which seems to be that accessibility-related 
GNOME cvs commits, (which in effect means all cvs commits), should be 
reviewed by accessibility engineers, seems unlikely to gain popular 
support - for one thing, there aren't enough accessibility engineers to 
review all the relevant changes, and for another, in general GNOME 
developers are not accustomed to having their code approved by outside 
reviewers, i.e. people who aren't official maintainers of their modules. 

In my opinion a more effective and palatable approach would be for more 
maintainers and distros to run accessibility regression tests such as 
those which Sun runs, there are available for public view at 
http://developer.gnome.org/projects/gap/sanity-testing/index.html

best regards,

- Bill

Jason White wrote:

> On Tue, 26 Apr 2005, Bill Haneman wrote:
>
>  
>
> > I'd be grateful for anything that distros or other maintainers can do to
> > help expedite getting these patches into gnome cvs.
> >    
>
> This raises the question of what projects such as Gnome should be doing to
> ensure that accessibility-related improvements are expeditiously and
> properly maintained and integrated. I'm not acquainted with the
> organizational structure of the Gnome project, but I assume there is a
> process whereby code is reviewed before it is committed to cvs.
>
> Perhaps someone involved with accessibility should be given cvs write
> access and authority to review and commit accessibility-related changes.
>
> Failing this, would there be a straightforward way of generating patches
> that just contain the accessibility-related changes, which distributors
> could be encouraged to apply? The ideal would be to use a revision control
> system that tracks Gnome cvs, contains the accessibility-relevant fixes
> and can generate patches without manual intervention. Another project I am
> familiar with, namely the xfs file system for Linux, effectively did this
> for quite a long time by maintaining a kernel tree in cvs and regularly
> generating patches that were uploaded to an ftp server. The problem was
> that some manual effort was needed to create the patch sets, and of course
> we want developers to concentrate on getting work done rather than
> having to compensate for delays in the process attributable to project
> organization.
>
> Also, I hope the OpenSolaris project, which is now reportedly in the
> process of defining its organizational structure, leads by example in
> according priority to accessibility and internationalization, while
> establishing a government model which enables improvements to be properly
> reviewed and efficiently integrated.
>
>