[orca-list] Pulseaudio and speech-dispatcher/gnome-speech in Ubuntu.

[Luke Yelavich]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Wed, Apr 16, 2008 at 09:46:35PM EST, Hynek Hanke wrote:
> Luke Yelavich napsal(a):
>> I don't quite understand. 
> The original problem was that installing Dispatcher and configuring  
> /etc/defaults/ to execute it will destroy all other audio in Hardy,  
> because PulseAudio won't be able to start. This problem is totally  
> independent of whether /etc/defaults/speech-dispatcher is set to "yes"  
> or "no".

Ok, I'm with you so far.

> It is not a problem of Speech Dispatcher and there is no way how I could  
> fix it as an upstream author. It is an inconsistency in the  
> distribution, because switching to Pulse Audio output and not providing  
> any means of audio output independent of gnome-session is simply broken  
> and this is one of its consequences.

Ok, things wouldn't work outside of a gnome session. However, Ubuntu's BrlTTY 
package is not built against speech-dispatcher, since speech-dispatcher is in a 
different part of the Ubuntu archive, and the packages in the archive where 
BrlTTY sits now, can not be built against packages where speech-dispatcher 
sits. As for yasr, well at this point I would think that users are capable of 
getting things set up to suit them, considering the way yasr works in the first 
place. I'd be happy to be proven wrong however.

> Your recomendation is to either turn off Pulse Audio (which is likely to  
> break a lot of things since it is the default in Ubuntu Hardy) or for  
> the user to do a lot of configuration and run it under the user session.  
> But in the later case, how will the user make BrlTTY on text console or  
> Speakup speak?

As stated, BrlTTY doesn't have speech-dispatcher support, and Ubuntu hardy 
doesn't have speakup. As for turning off pulseaudio, other applications 
generally auto-detect the best output to use, or, if pulse doesn't work, they 
fall back to using Alsa.

> The system wide speech service must be available, but I see that it  
> won't work (at least not without a lot of configuration) in Hardy now.

As I think I've stated, things were done in such a way, because from reading 
the user posts on this list and other lists, users appeared to be happy to 
configure speech-dispatcher to suit their needs. 

>> The way I see things currently, given that my upload is allowed, is  
>> this. The system is installed with gnome-speech. The user wishes to  
>> switch to speech-dispatcher, so they install it. 
> Wishes to switch? You seem to be speaking from the point of view of  
> Orca, but what about BrlTTY, Speakup, speechd-el, Yasr? They can't work  
> with gnome-speech.

No of course they can't, but I am thinking of what is most commonly used. I 
haven't heard of users using yasr/speechd-el on Ubuntu, and if there are, and 
they think things could have been done better, it would have been nice to hear 
from them to set things right. Same with speakup, and for people who may have 
rebuilt BrlTTY.

>> Then they can then go about configuring speech-dispatcher the way they  
>> want.
> Yes, but they should be allowed to start from a state of things that  
> works, not from one which is broken.

Again, what I've done is from what I've perceived from the community. People 
need to really speak up if they think things could be done a lot better. It 
seems that some of the time, I have to guess what users may want. Again, if 
I've totally misread the goings on of this community, I would be happy to be 
put straight.

>>> Can't we at least change the configuration of the output modules from 
>>>  "alsa" to "pulse"
>>>     Yes, I could do that.
>>>     
> Great. But this is only meaningful if also:
>>> and start a Pulse Audio daemon under the speech-dispatcher user from  
>>> /etc/init.d/speech-dispatcher?
>>>     
>> No, we can not. PulseAudio will never run system wide, or even specifically 
>> for another daemon user by default, particularly at this late stage. There 
>> is just over a week
>> till hardy is released, I'm pushing it as it is.   
>
> What do you mean by ,,will never run for another daemon user by  
> default''? Is there any technical reason or is this your decision?  
> Because that would fix it, it is obvious and it is necessary.

Put it this way. If I was to do it the way you suggest, I wouldn't be popular 
with my fellow developers, and it would make things even more messy. Since 
pulseaudio uses the sound device directly, i.e not through dmix, this would 
totally prevent users running pulseaudio as a user, unless they had another, 
unused sound card. And using pulseaudio through dmix presents another pile of 
problems, which are to be avoided at all costs. General applications wouldn't 
then be able to use alsa.

> I understand this is a late release stage. But it is simply a serious  
> issue and we have recently seen many people complaining, so I'm trying  
> to discuss these things to propose a solution to fix that problem and I  
> believe we should search for ways how to do it.

I understand your position and point of view, and I hope this email has 
attempted to clear up mine. So we have a few days to make a decision. I've been 
given permission to upload, so this is the chance to get everything in, and get 
it right.

Thanks for your comments.

Luke
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFIBfEHjVefwtBjIM4RAsicAKCiqEOAlyrVyGjvXM9d7DwiLzKUfACgwyyU
xlzzOXtCBxDzIVS+fZXMDAQ=
=C5Qh
-----END PGP SIGNATURE-----