This is a draft.

Speakup

Current problems:

  1. BrlTTY can't work with session instances of Speech Dispatcher, thus forces the user to deal with the problems of system-wide Speech Dispatcher. Most importantly, it can't work with Pulse Audio, which makes it broken on Ubuntu, Fedora and Mandriva unless the user severly reconfigures his system.
  2. Can't be started by the user, needs to be started via etc/init.d

Proposed solution:

  1. /dev/softsynth and related devices need to be group-readable-writable via udev
  2. speechd-up will be started via:
    1. idle instance: etc/init.d
    2. user-session instance: via accessible login (which we do not have...)
  3. speechd-up needs to free /dev/softsynth when not in active session

BrlTTY

Current problems:

  1. BrlTTY can't work with session instances of Speech Dispatcher, consequences as for Speakup.
  2. BrlTTY needs to be manually configured for the given braille display
  3. Can't be started by the user, needs to be started via etc/init.d

Proposed solution:

  1. should discover devices through udev (and spare users the configuration)
  2. /dev/vcsa group readable via udev (for now)
  3. free devices when not in active session
  4. autostarted in the same way as Speakup