On 12/24/2014 6:48 PM, Adam Borowski wrote:
I think it's better to think about what do we get to gain and what to lose
from pulseaudio.
Gains:
* ability to reroute streams mid-run (like, speakers->headphones), but only
if they are connected to physically separate sound cards
* a common way to direct streams (without, you need to read man pages and
learn obscure arguments that differ per program -- if a program has the
ability to use a non-default output at all)
* no need for manual configuration of certain outputs, like bluetooth
headphones
* can do basic configuration of certain commonish scenarios (like 5.1) for
you
Have you ever tried to explain how to configure ALSA manually to a non
programmer? PA might be a PITA, but it does keep them from having to do
that in the vast majority of cases. If you think PA is annoying, wait
until you have to manually configure a mixer because ALSA has locked the
card to one stream. There is also the fact that lot of higher level
software does not function properly when making direct calls to ALSA for
the same reason: locking.
Problems:
* new breakage (won't work if ALSA doesn't, is likely to fail on its own)
* buggy like hell:
Granted. Been there, done that. But ALSA is hardly perfect either.
I've found version 4 to be more behaved.
* likes to take 100% of a CPU core for no obvious reasons
Older versions, yes. There is also the fact that is more than just one
layer to the audio stack, you can't always blame PA. Jack is known for
the same issue. Either can go south if it is not setup correctly.
In other words, it does significantly increase user friendliness while
providing little actually new functionality.
Not really, PA provides mixing for individual applications, using just
ALSA does not.
I agree absolutely that there a very sanguine reasons for dropping PA,
I just think that before we throw out the "baby with the bathwater" we
should consider what that means. Like it or not, there is an inordinate
amount of software that depends on PA, and we don't have a drop in
replacement.
_______________________________________________
Dng mailing list
[email protected]
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng
