Subject: Re: [linux-audio-dev] News about sequencers (not my own though!)
From: Paul Barton-Davis (pbd_AT_Op.Net)
Date: ti tammi 25 2000 - 22:52:07 EST
>I'm thinking this is like bytecode except not smashed flat: you want
>to have the bulk of your inner loop's work happen inside a library
>call (an opcode, here), because otherwise you lose on overhead. In
>Csound this is entangled with the fact that when you have to program
>something there's no opcode for, you probably have to set kr=ar too.
^^
??
sr ?
"inner loop" has two meanings here. one is the inner loop, typically
seen in Csound opcodes:
int nsmps = ksmps;
while (--nsmps) {
... do something ...
}
and the other is the "inner loop" of the DSP emulation (in Csound,
this can be found in the function kperf(); in Quasimodo, its the
function DSP::do_work ()).
>The sort of programming I'm particularly interested in is musical DSP,
>i.e. writing new opcodes using a high-level language. (If I never do
>it in C again it'll be too soon.) I don't imagine there's any way to
>have an efficient interpreter for such a language.
I'd like to claim that quasimodo will be exactly that: an efficient
interpreter for a high-level language for DSP. The language primitives
will be the existing opcodes (mostly from Csound, but we have some new
ones already), plus new ones written in C++, combined with C-like
arithmetic, logical, boolean and unary operators.
What the form of the first chosen language will be is up for
grabs. I've stared at SAOL and have concluded that it is
undesirable. This is mostly because of its propagation of Barry
Vercoe's mistake that "rate" is a language-level construct. I can cc:
you or the list my letter to Eric Scheirer if you're interested.
I am open to any suggestions. Using a new language is mostly a case of
writing a bison/flex parser/lexer for it, and plugging it in. The
framework for this is not all there, but I have designed what is there
with the necessary extensions in mind. If anyone has good ideas for
either a replacement for the Csound orc language, or the Csound sco
language (i am predisposed to NeXT's Scorefile for the latter).
--p
This archive was generated by hypermail 2b28 : pe maalis 10 2000 - 07:23:27 EST