1 files changed, 73 insertions, 6 deletions

diff --git a/src/libs/engine/Connection.cpp b/src/libs/engine/Connection.cpp
index 1f5e0d21..992496c2 100644
--- a/src/libs/engine/Connection.cpp
+++ b/src/libs/engine/Connection.cpp
@@ -19,6 +19,8 @@
 #include "util.h"
 #include "Node.h"
 #include "Port.h"
+#include "BufferFactory.h"
+#include "AudioBuffer.h"
 
 namespace Ingen {
 
@@ -28,17 +30,82 @@ namespace Ingen {
  * This handles both polyphonic and monophonic nodes, transparently to the 
  * user (InputPort).
  */
-Connection::Connection(Port* const src_port, Port* const dst_port)
-: _src_port(src_port),
-  _dst_port(dst_port),
-  _pending_disconnection(false)
+Connection::Connection(Port* src_port, Port* dst_port)
+	: _src_port(src_port)
+	, _dst_port(dst_port)
+	, _local_buffer(NULL)
+	, _buffer_size(dst_port->buffer_size())
+	, _must_mix(src_port->poly() != dst_port->poly())
+	, _pending_disconnection(false)
 {
-	assert(src_port != NULL);
-	assert(dst_port != NULL);
+	assert(src_port);
+	assert(dst_port);
+	assert(src_port->type() == dst_port->type());
 
 	assert((src_port->parent_node()->poly() == dst_port->parent_node()->poly())
 		|| (src_port->parent_node()->poly() == 1 || dst_port->parent_node()->poly() == 1));
+
+	if (_must_mix)
+		_local_buffer = BufferFactory::create(dst_port->type(), dst_port->buffer(0)->size());
+}
+
+
+Connection::~Connection()
+{
+	delete _local_buffer;
 }
+	
+
+void
+Connection::set_buffer_size(size_t size)
+{
+	if (_must_mix) {
+		assert(_local_buffer);
+		delete _local_buffer;
+
+		_local_buffer = BufferFactory::create(_dst_port->type(), _dst_port->buffer(0)->size());
+	}
+	
+	_buffer_size = size;
+}
+
+
+void
+Connection::process(SampleCount nframes, FrameTime start, FrameTime end)
+{
+	// FIXME: nframes parameter not used
+	assert(_buffer_size == 1 || _buffer_size == nframes);
+
+	/* Thought:  A poly output port can be connected to multiple mono input
+	 * ports, which means this mix down would have to happen many times.
+	 * Adding a method to OutputPort that mixes down all it's outputs into
+	 * a buffer (if it hasn't been done already this cycle) and returns that
+	 * would avoid having to mix multiple times.  Probably not a very common
+	 * case, but it would be faster anyway. */
+	
+	// FIXME: Implement MIDI mixing
+	
+	if (_must_mix) {
+		assert(type() == DataType::FLOAT);
+
+		AudioBuffer* mix_buf = (AudioBuffer*)_local_buffer;
+
+		//cerr << "Mixing " << src_port()->buffer(0)->data()
+		//	<< " -> " << _local_buffer->data() << endl;
+
+		mix_buf->copy((AudioBuffer*)src_port()->buffer(0), 0, _buffer_size-1);
+	
+		// Mix all the source's voices down into local buffer starting at the second
+		// voice (buffer is already set to first voice above)
+		for (size_t j=1; j < src_port()->poly(); ++j)
+			mix_buf->accumulate((AudioBuffer*)src_port()->buffer(j), 0, _buffer_size-1);
+
+		// Scale the buffer down.
+		if (src_port()->poly() > 1)
+			mix_buf->scale(1.0f/(float)src_port()->poly(), 0, _buffer_size-1);
+	}
+}
+
 
 } // namespace Ingen