Re: [g-a-devel] What's the Status of Audio in GDM?




Chuck:

On Wed, May 07, 2008 at 10:13:06AM -0400, Janina Sajka wrote:
What is the plan for SoundOnLogin=true ? Is it coming back? Should I
just be patient?

I don't know about that option, however I do have sound working on my GDM login screen. Orca starts up automatically.

Running orca (or other AT programs) by default is good for single-user
machines.  However, it is not a great solution for multi-user machines.
There really needs to be a way to launch the AT programs, including
keybinding and mouse gesture launching.

And, what's the thought on providing audio device support at this stage?
Under the default pulseaudio, configuration the audio device isn't
available until after the login succeeds.

I didn't do anything special to enable audio device support or pulseaudio on GDM. It just worked for me by default after installing. All I had to do was click on the accessibility icon in the lower left corner of the login screen to turn on "Hear text aloud". If there was a shortcut-key for this option would that be adequate?

If you mean a shortcut key to launch "Hear text aloud", then this would
be adequate.  However, a shortcut key to launch the dialog where they
can select "Hear text aloud", and then having to navigate the dialog
to launch this choice would be more difficult without feedback.

It would probably be best to come up with a single solution that allows
people to launch Orca in both GDM and the user session, rather than
implementing a GDM specific solution.

Lastly, let me note that there's a security dimension here that can only
be solved by clearing up a11y login. Without appropriate At support,
it's not all that hard to get out of sync with the GDM dialog and put
one's password where the login ID belongs.

The new GDM greeter by default lists all the local users on the system, and you select the user's name from the list to log in as that user. Then you type the password. So there won't usually be a need to type your login ID, unless this is the first time you are logging into a system with remote users (NIS, LDAP, etc.) in which case your user name might not be in the list yet. In that case, you have to click "Other..." and then type the login ID.

Navigating a list is more complicated for blind users, since they
have to process each line of information.  Text-to-speech engines
may not do a good job of reading arbitrary user names.  On multi-user
machines, there could be many users in the list, making it time
consuming to navigate.  A blind user would likely prefer to simply enter
their username rather than having to navigate a list.   Having to
navigate to "Other" is perhaps not too bad, as long as the blind
user can get to it quickly and without first having to parse through
a bunch of user names.  Though the meaning of the label "Other" isn't
really isn't so clear.  A blind user might not understand what it
means at first.

I did notice that the text that Orca is reading is less than helpful. First of all, it says nothing at all when the login window is ready to be used. It doesn't start reading until I click something on the login widget.

Blind users, who mainly use Orca, typically do not use a mouse at all.
Does hitting a key (tab or enter) make Orca come alive?

Second, after clicking my user's full name in the list, it says "Password:" and sometimes it reads each key of my password as I type it instead of saying "star". I can't reproduce this problem all the time, however. I think there are some race conditions between when GDM starts up pulseaudio and Orca, because I have heard the "Welcome to Orca" occasionally when the login window is ready.

Finally, I noticed that after logging in, Orca says "Welcome to Orca" but then the process is killed or dies for some reason, and the rest of my session no longer has the screen reader enabled, even though I have it enabled in my session preferences. Logging out doesn't fix it either, and either Orca or pulseaudio (not sure which) no longer runs on the GDM screen. Rebooting seems to fix things.

So it seems that the support for accessibility, including audio and the screen reader on the login screen exists, it is just that there are many bugs that need to be ironed out.

It would also be good to test how well a11y works if two users are
logged into separate sessions on the same machine (perhaps one
via the console and another via XDMCP).  I don't think the rewrite
yet supports XDMCP, but when it does this would be a useful test
case.

Brian


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]