Linux WISH LIST
David Cabrero Souto
cabrero at udc.es
Fri Dec 12 12:20:14 GMT 2008
El 12/12/2008, a las 0:05, mike escribió:
> I agree, a talking bootloader would be nice. I don't know how this
> could be done with software speech though? Maybe another Linux
> kernel could load and then offer speech to pick from the ubuntu
> menu. I realize booting would take longer, but it would be worth it
> to have the accessibility.
> We also need a good soleution for e-mail.Mike.
We, a research group at the University of A Coruña, Spain, have been
developing a small toy implementation based on the idea of Mike. It's
far from complete, neither usable for final users, but we have been
able to validate positively the idea. I will try to explain what we
did and what are our conclusions.
The mechanics of the system are:
1. The system boots directly into a linux kernel with a special initrd
disk image. That initrd launches a boot menu with speech synthesys and
2. Once the user selects the OS to boot we use the kexec service of
the kernel to start up the new operating system. The main advantage of
using kexec is that does its work whithout rebooting the machine. This
saves tons of time of the booting process. In our tests, kexec was
able even to start grub4dos, and grub4dos started a ms windows OS.
Always whithout any reboot.
3. Whenever the user chooses to boot ubuntu, it starts up using a
customized initrd that adds sound feedback to the boot progress bar.
The feedback is just a ping-like sound every time the slide reaches
one ending of the progress bar.
And our conclusions are:
1. We didn't implemented everything, but the prototype was functional
and the resulting system was viable. We think that actually
implementing the systems is worth the effort, but there are some
problems to solve ...
2. The initial initrd is quite big because it must contain the
- The kernel modules for the sound card drivers
- The alsa subsytem, containing customized config files and librarys
- An application that runs a grub-like menu in order to allow the
user to choose the OS she wants to boot
- The espeak speech synthetizer plus the c++ library it depends on.
- The brltty for braille support.
3. It is necessary to automate the customization of the linux OS
initrd in order to add sound feedback to the progress bar. This
customization must be done in a per-distribution basis, in other
words, what we will do for ubuntu may not be usefull for other distros.
4. It is necessary to automate the update of the boot menu list every
time the package manager adds a new kernel. I think it will be a lot
easier if the boot menu could read directly the grub menu config file.
At present we are not working in this project due to the lack of human
resources, but i'd like to continue as soon as possible.
I hope this information would be usefull. Best regards.
More information about the Ubuntu-accessibility