Wouldn't it be nice if plugin presets had a description/comment?
[ardour.git] / libs / ardour / unknown_processor.cc
old mode 100755 (executable)
new mode 100644 (file)
index 91c9bc9..3a5ad42
 
 */
 
+#include "ardour/audio_buffer.h"
 #include "ardour/unknown_processor.h"
 
+#include "pbd/i18n.h"
+
 using namespace std;
 using namespace ARDOUR;
 
 UnknownProcessor::UnknownProcessor (Session& s, XMLNode const & state)
        : Processor (s, "")
        , _state (state)
+       , have_ioconfig (false)
+       , saved_input (0)
+       , saved_output (0)
 {
        XMLProperty const * prop = state.property (X_("name"));
        if (prop) {
                set_name (prop->value ());
+               _display_to_user = true;
+       }
+
+       int have_io = 0;
+       XMLNodeList kids = state.children ();
+       for (XMLNodeIterator i = kids.begin(); i != kids.end(); ++i) {
+               if ((*i)->name() == X_("ConfiguredInput")) {
+                       have_io |= 1;
+                       saved_input = new ChanCount(**i);
+               }
+               if ((*i)->name() == X_("ConfiguredOutput")) {
+                       have_io |= 2;
+                       saved_output = new ChanCount(**i);
+               }
        }
+       have_ioconfig = (have_io == 3);
+}
+
+UnknownProcessor::~UnknownProcessor () {
+       delete saved_input;;
+       delete saved_output;
 }
 
 XMLNode &
-UnknownProcessor::state (bool)
+UnknownProcessor::state ()
 {
        return *(new XMLNode (_state));
 }
 
+bool
+UnknownProcessor::can_support_io_configuration (const ChanCount &in, ChanCount & out) {
+       if (have_ioconfig && in == *saved_input) {
+               out = *saved_output;
+               return true;
+       } else if (!have_ioconfig) {
+               /* pass for old sessions.
+                *
+                * session load assumes processor config succeeds.
+                * if initial configuration fails, processors downstream
+                * remain unconfigured and at least the panner will assert/segfault.
+                *
+                * This may still result in impossible setup, however
+                * Route::configure_processors_unlocked() ignores configure_io() return value
+                * in the inner loop and configures all available processors.
+                *
+                * It can still lead to segfaults IFF the track has no inputs and this is a
+                * generator (processor_max_streams will be zero).
+                */
+               PBD::warning << _("Using plugin-stub with unknown i/o configuration for: ") << name() << endmsg;
+#if 0
+               /* No output channels are fine (or should be, there may be edge-cases with e.g sends).
+                *
+                * Discussion needed.
+                *
+                * This is the safer option (no audio output, no possible damage)
+                * and the way to go in the long run.
+                * An even better solution is to disable the route if there are missing plugins
+                * and/or impossible configurations.
+                *
+                * Currently no output channels results in awkward GUI route display and also
+                * breaks semantics in mixbus (which assumes that the route has channels required
+                * for the always present mixer-strip plugin).
+                */
+               out = ChanCount ();
+#else
+               out = in;
+#endif
+               return true;
+       } else {
+               PBD::error << _("Using plugin-stub with mismatching i/o configuration for: ") << name() << endmsg;
+               out = in;
+       }
+       return true;
+}
+
+void
+UnknownProcessor::run (BufferSet& bufs, samplepos_t /*start_sample*/, samplepos_t /*end_sample*/, double /*speed*/, pframes_t nframes, bool)
+{
+       if (!have_ioconfig) {
+               return;
+       }
+       // silence excess output buffers
+       for (uint32_t i = saved_input->n_audio(); i < saved_output->n_audio(); ++i) {
+               bufs.get_audio (i).silence (nframes);
+       }
+}