Subject: [linux-audio-dev] laaga: withdrawal/amendment of automated mixing proposal
From: Paul Davis (pbd_AT_Op.Net)
Date: Fri Jun 15 2001 - 23:41:38 EEST
i have just realized a fundamental problem with the automated mixing
proposal that i made a few weeks back.
the mixing was to be done by the engine, but it has to be done
immediately before executing the client whose inputs might be mixed.
the problem is that for out-of-process clients, we don't want to
switch back and forth between the clients and the engine, we want to
switch from one to the next and then back the engine only when we have
to. this means that we can't mix at the right time, which means that
we can't mix at all.
instead, i would suggest we adopt richard's idea of using "hidden"
mixing nodes which would do the mixing for the client. the engine
would create them on demand, and insert them at the appropriate point
in the flow graph. these nodes would be internal (i.e. not out of
process), and thus the cost is small.
an example: Client #1 already has one connection to its (audio) input
port. The user creates a new one. The engine creates a 2 input/1 output mix
node, reconnects the original connection to the first port, the new
connection to the second, then feeds the output to Client #1.
i don't know how to handle this in terms of GUI displays, but my
instinct suggests that it should not be a problem.
its either that or we drop automated mixing completely, which would be
a shame, since it means no Many-to-One connections.
or we do even more context switches, which doesn't seem too good of an
idea.
Your Comments Please.
--p
This archive was generated by hypermail 2b28 : Fri Jun 15 2001 - 23:38:01 EEST