Subject: Re: [linux-audio-dev] Remote control is important
From: Paul Barton-Davis (pbd_AT_Op.Net)
Date: Sun Nov 19 2000 - 00:34:34 EET
>squeeze out the most DSP I can, so if I can get the laptop to take
>care of more than just the drawing requests that would be nice.
the GUI work that is *not* related to drawing requests consumes an
almost immeasurable fraction of CPU power. its the drawing stuff and
the interaction with the video subsystem that is worth moving; i don't
think that moving most of GTK, for example, helps much at all, whereas
moving the X server cuts out a notable load.
>Especially if I'm doing something like, say, interactive algorithmic
>music where the part of the system generating events is quite complex.
yeah, now this part i can see the point of.
>3) libsig++ when it's network transparent ... don't know much about it
>except reading PBD's posts. Just to clarify: Does this mean I'll be
>able to run, say, Quasimodo's engine on one box while its UI is on
>another box?
you can already do that (modulo any current compile-time issues). its
called server-quasimodo, and it has a command-line parser which can be
controlled by a (G)UI on another system. stephane conversey was
responsible for most of this, and he had a Python-GTK GUI that he ran
on one machine while server-quasimodo ran on some other system.
but this has nothing to do with libsigc++ being thread and/or network
transparent.
>4) something much like a unix named pipe, only working over TCP/IP
>(e.g. over ethernet). That would be a LOT of fun for even a
>know-nothing like me. And it wouldn't care what kind of data you were
>sending - MIDI, ASCII, audio data,
its called a socket. it takes a couple of extra calls to set up
compared to a pipe. see "man socket(2)".
--p
This archive was generated by hypermail 2b28 : Sun Nov 19 2000 - 01:09:53 EET