Dumping errors to stderr only is not very useful. Particularly not
on Windows and MacOS.
Even though a user may not be able to address the issue, this
can lead to better reports vs just printing "corrupt state".
Otherwise this will lead to a corrupt state:
ERROR: Session: XMLNode describing a AudioRegion references an unknown source id
ERROR: Session: cannot create Region from XML description. Can not load state for region
ERROR: Playlist: cannot create region from XML
and a track without playlist is created, resulting in a later crash.
Eventually SessionPlaylists::load() needs to handle this gracefully,
but this should help catch cases causing the actual issue.
~StatefulDiffCommand() may trigger UndoTransaction::command_death()
which may delete the StatefulDiffCommand() that's just being destroyed.
This depends on the signal-connection order, which is undefined.
In any case when a shared_ptr<> object is being destroyed it means
that all references to it are already gone. There's no need to
emit drop_references from the d'tor.
On session re-load only automation lanes with events were displayed,
regardless of visibility state. This allowed for inconsistent
state (menu showed them as visible even if they were not).
This fixes a case when deleting a plugin, deletes all automation
undo/redo events:
<UndoTransaction name="add automation event">
<MementoCommand type-name="ARDOUR::AutomationList">
...
`delete this;` calls the d'tor which emits drop_references(),
that leads to UndoTransaction::command_death() destroying the
object, whichh causes a double free.
There are various ways to cancel a dialog. Only checking for
RESPONSE_CANCEL is not sufficient. e.g. Esc causes a delete-event.
* Gtk::RESPONSE_CLOSE
* Gtk::RESPONSE_REJECT
* Gtk::RESPONSE_DELETE_EVENT
* Gtk::RESPONSE_CANCEL
* Gtk::RESPONSE_NO
Among others this fixes "Clicking session > open,
then hitting ESC opens the currently selected folder and session"
set_auto_punch_location() is a NO-OP when there's no punch range.
This disconnects `punch_connections`, clears session-events,
and emits auto_punch_location_changed().
The previous approach failed in case where PluginInsert
uses no-inplace buffers with a linear map.
Since buffers are replicated up to a total of number of
all (inputs + outputs), the number of output buffers
could not be determined. There was insufficient information
using the I/O map alone.
With a known number of outputs processing and applying
the i/o map is also a lot easier and faster.
This break the API of process_map().
(valgrind trace, line-numbers from mixbus+6.0-190-g0ec6bc35a)
This may happen initially for unconnected graph nodes,
e.g. Foldback Busses from ARDOUR::Session::post_engine_init().
==29797== Conditional jump or move depends on uninitialised value(s)
==29797== at 0x6167D3F: trace_terminal(boost::shared_ptr<ARDOUR::Route>, boost::shared_ptr<ARDOUR::Route>, bool) (session.cc:2174)
==29797== by 0x6167D99: trace_terminal(boost::shared_ptr<ARDOUR::Route>, boost::shared_ptr<ARDOUR::Route>, bool) (session.cc:2174)
==29797== by 0x6167D99: trace_terminal(boost::shared_ptr<ARDOUR::Route>, boost::shared_ptr<ARDOUR::Route>, bool) (session.cc:2174)
==29797== by 0x616890D: ARDOUR::Session::resort_routes_using(boost::shared_ptr<std::__cxx11::list<boost::shared_ptr<ARDOUR::Route>, std::allocator<boost::shared_ptr<ARDOUR::Route> > > >) (session.cc:2289)
When a playlist is deleted and drops_references(), any
undo/redo StatefulDiffCommand referncing playlist invoke
Destructible::drop_references() of the Command.
This leads to command_death(). As opposed to UndoTransaction::clear()
the StatefulDiffCommand was not destroyed.
In case of playlists StatefulDiffCommand::_changes contains
PBD::SequenceProperty<std::list<boost::shared_ptr<Region> > >
and shared pointer reference of the playlist regions were kept
indefinitely.
This fixes the following scenario:
New session, import an file, delete the created track,
clean up unused sources (delete unused playlists)[, quit].
A reference to the imported region was kept, because of the
playlist's undo command (insert region). Yet the source file
was deleted.
PS. Most playlist changes are accompanied by GUI zoom/selection
MementoCommands. Those are currently never directly dropped.
command_death() leaves those in place.
Various operations clear the history (e.g. cleanup). In that
case the GUI correctly had an empty Undo/Redo history, but the
file on disk was left in place.
Next session load restored the old, incorrect Undo/Redo history.
This type of MIDI port fetches all of its data from inside ::cycle_start(),
and delivers it to a FIFO connected to another thread (typically a
control surface).
Unlike regular MidiPorts, which will be read from inside a Session::process()
call, these ports will read their data once per AudioEngine::process() cycle.
They therefore cannot use MidiPort::get_midi_buffer() which scales and adjusts
event timestamps as if the data is being accessed from within Session::process().
It is still an open question whether or not AsyncMIDIPort::cycle_start() should
still scale event timestamps by speed. In some respects it seems more appropriate
to do so, and the reading thread (e.g. a control surface) doesn't care about
the "nframes" limit on timestamps that exists for calls within a Session::process()
tree. For now, leave the timestamps unscaled by speed.
