[linux-audio-dev] What parts of Linux audio simply suck ?

Christoph Eckert mchristoph.eckert at t-online.de
Sun Jun 19 20:41:25 UTC 2005


OK,


> And your favorite is... ?

you asked for it :) .

Some things which I'd really appreciate:

* Applications could remember tha last used MIDI and JACK 
connections as well as the last used file and resetup 
itselves when getting restarted. Applications could write 
these infos to the config file as soon as connections are 
made and a file gets saved/opened, so a crashing application 
can still remember these things. This is something which fits 
to Dave's request and can reduce the need of LASH

* When JACK gets restarted it would be great if running JACK 
applications wouldn't need to be restarted. Perhaps it would 
be possible to teach them to bring an informational message 
to the user that the connection to JACK has been lost (some 
already do) and to reconnect automatically as soon as JACK 
gets available again. I do not know how difficult this is 
from the technical view but I guess it can be done. This also 
fits the first issue and can also reduce the need for LASH

* Applications could complain via an alert box as soon as the 
needed audio subsystem isn't available, e.g. JACK isn't 
running or the ALSA device is blocked. ams simply doesn't 
appear on the screen but is waiting in the RAM until the ALSA 
device gets available. As soon the blocking application quits 
ams appears on the screen. This is not only an ams issue, but 
I made this experience with ams so it's an example

* Applications which can output sound to various audio 
subsystems could offer an autodetection mechanism.
Example: Application Y can output sound via ALSA or JACK. 
Invoking it without any parameter makes it running on top of 
ALSA. To make it an JACK application, it needs the parameter 
--jack.
I'd prefer the following behaviour:
If it is invoked without any parameter, it looks for JACK. If 
JACK is running, it behaves as a JACK client. If not, it 
tries to use ALSA. The autodetection can be avoided by 
explicitely telling which system to use.

Examples:

Y --jack => makes it a JACK applicaton
Y --alsa => makes it an ALSA application
Y           => does autodetection






Best regards


    ce




More information about the Linux-audio-dev mailing list