when you evaluate a behavior is important...
Posted: Mon Aug 12, 2013 2:21 am
This is a little note for anyone just getting started in Nyquist. Like me. Perhaps this will save somebody some confusion and astonishment.
Consider the following code evaluated in Nyquist:
You might expect snd-extent to produce the same results for the two cases: we're starting with the same envelope function (pwl 1 1 1) and stretching it by the same amount (stretch 4.0 <expression>). Let's see what happens:
Oho! The extent of envelope-var ends at 1, the extent of (envelope-fun) ends at 4.
An expert should correct me if I'm wrong about this, but the crucial difference is when (pwl 1 1 1) is evaluated. For envelope-var, the pwl function is evaluated BEFORE stretch is called. In envelope-fun, pwl is evaluated within the context of the stretch. Behind the scenes, pwl looks at the global *warp* variable to determine its start and stop times (more accurately, the mapping between local and global times). So when pwl is evaluated to set envelope-var, stretch is not in effect, so the result isn't subjected to stretching. On the other hand, when (envelope-fun) calls pwl, stretching is already in effect, so the result *is* properly stretched.
Moral: make sure you don't evaluate a behavior until you're in the correct evaluation context.
(uh, did I get that right?)
Consider the following code evaluated in Nyquist:
Code: Select all
(setq envelope-var (pwl 1 1 1))
(display "envelope-var" (snd-extent (stretch 4.0 envelope-var) 1000000))
(defun envelope-fun () (pwl 1 1 1))
(display "envelope-fun" (snd-extent (stretch 4.0 (envelope-fun)) 1000000))
Code: Select all
envelope-var : (SND-EXTENT (STRETCH 4 ENVELOPE-VAR) 1000000) = (0 1)
envelope-fun : (SND-EXTENT (STRETCH 4 (ENVELOPE-FUN)) 1000000) = (0 4)
T
An expert should correct me if I'm wrong about this, but the crucial difference is when (pwl 1 1 1) is evaluated. For envelope-var, the pwl function is evaluated BEFORE stretch is called. In envelope-fun, pwl is evaluated within the context of the stretch. Behind the scenes, pwl looks at the global *warp* variable to determine its start and stop times (more accurately, the mapping between local and global times). So when pwl is evaluated to set envelope-var, stretch is not in effect, so the result isn't subjected to stretching. On the other hand, when (envelope-fun) calls pwl, stretching is already in effect, so the result *is* properly stretched.
Moral: make sure you don't evaluate a behavior until you're in the correct evaluation context.
(uh, did I get that right?)
