running speech-dispatcher-0.7.1 in system mode

[Trevor Saunders]

HI,

On Wed, Oct 06, 2010 at 03:58:02PM -0500, William Hubbs wrote:
> Hi Steve,
> 
> On Wed, Oct 06, 2010 at 12:23:22PM -0700, Steve Holmes wrote:
> > In following this thread, I have to say one thing that hasn't really
> > been addressed is kernel based solutions like Speakup.  The only way I
> > know if that one could use Speakup with Speech Dispatcher is through
> > speechd_up and AFAIK, that would require Speech Dispatcher to run as a
> > system service.  So we need to keep that functionality in place.  
> 
> Actually that's not quite true.  The only reason speechd-up runs as root
> is so that it can open the softsynth device.  Other than that,
> communication with sd doesn't necessarily have to be over an internet
> socket, and sd doesn't necessarily have to run as root.
> 
> I'm not to the point where I can check this yet, but here is the way I'm
> thinking it can work:
> 
> 1.  speechd-up is started as root.  This would be the only thing that
> would need a system startup script.
> 2.  speechd-up opens /dev/softsynth
> 3.  speechd-up drops privileges to some user other than root.
> 4.  speechd-up now opens a connection to speech dispatcher, which will
> cause speech dispatcher to autospawn.
> 
> What do you think?

I'm not convinced you can make this work with pulse, which imho is the
only real need for anything to be in a session.

 In dealing with pulse your problem is that pulse only wants to let the
 user it believes is sitting in front of the machine use the audio
 device.  SO speech-dispatcher has to run as your user.  Then speechd-up
 can either run as a system user or as your user.  If it is to run as
 you it has to know what user you are, or for speechd-up to run as a
 system user it has to know where to contact speech dispatcher.  I gues
 you could solve these problems but there's no nice fairly automatic way
 to do it, and in the case of speechd-up running as a system user its
 simpler if speech-dispatcher is also running system wide, since it can
 know a default address instead of having to have it configured.  On the
 other hand in any world were the audio system just uses the audio group
 to decide who can play sound this idea should work so long as the user
 running speech dispatcher is in the audio group.

 The easiest solution for all this all would be if the pulse people
 fixed the bug where root can't produce audio since root should always
 be able to do so.  It would be nicest if root could give that ability
 to a group of system users so that peech dispatcher wouldn't have to
 run as root, but running as root is fine if root can play sound.

 The idea of a speech dispatcher audio agent thing  that just recieves
 audio and plays it isn't bad either, it adds a process and maybe a little
 latentcy, but it should be tolerable especially if its started as root
 with a low nice value.

did you ever play with the udev idea we discussed a while back?

Trev

> 
> William
> 



> _______________________________________________
> Speechd mailing list
> Speechd at lists.freebsoft.org
> http://lists.freebsoft.org/mailman/listinfo/speechd

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: 
<http://lists.freebsoft.org/pipermail/speechd/attachments/20101006/d1957d68/attachment-0001.pgp>