John ffitch  wrote:
> I have seen a report that the change made to delayr and delayw to give
> better nesting of delay lines means that the well-known xanadu
> orchestra fails.  Unfortunately I cannot remember exactly what was
> done, or when.  If anyone can enlighten me I would be grateful (a
> failure of my ChangeLog i am afraid).  Anyway, I am not sure what to
> do about this.
>   Thoughts?
> ==John ffitch

Hello list, hello John!
Unfortunately, you are right.
It was me who wanted and who programmed the changes. They were meant to
improve interleaving (I wouldn't say nesting) delayr/delayw pairs. But I
had not thought that it might make sense to access the delay line data
from _outside_ the delayr/delayw pair. Seemingly it does, as Xanadu.orc
shows. I'm sorry about that.
With my changes, there is a mechanism to associate a deltap(i) opcode
with the right delayw opcode at init-time. It keeps track which delay
line is the current one by validating a delay line reference with delayr
and invalidating it with delayw. After the last delayw (and before the
first delayr) there is no current delay line, and trying to access one
is interpreted as error. That's what breaks the Xanadu orchestra.
I'm trying to find a solution which makes Xanadu.orc and similar ones
work as they did before without sacrificing the new possibilities.
Please, give me a little time before you decide to undo the changes.

So long,