Here are belated replies to some of Larry's querys --

1. I understand that DSP signal inlets and outlets are in the code, but

according to MSP aren't quite working right yet. What problems are there (should I try using them or not)

They're working but inefficient (the data gets copied TWICE on the way in and once on the way out as I recall.) I'll improve the situation later. For the moment, it might be more efficient to use send~/receive~.

2. In the "stuff" directory, at least one of the examples uses a

technique of patching two different signals to the same inlet, and the effect seems to be that the two signals get added at the inlet. The receiving object in this case is a "cos~". Is this technique valid in general? If so, what part of the PD source code implements this?

Yes, signal inlets to Pd objects are additive.

I notice in the Ircam Max docs, there is a description of an "expr"

object. Will this be implemented soon?

A student of mine (Travis Newhouse) is working on it and seems likely to get it done this Spring. I also want to provide an "expr~" object...

As to "linear languages" I have a daydream of providing a Forth interpreter. Failing that, it might be that writing C code is the most effective way to do sequential stuff in Pd...

Miller