It turned out that 'boost::intrusive::list_base_hook<>' won't compile if its parent class is declared using '__declspec(dllexport)' - so rather than exporting each entire class, let's use the alternative approach and export the various class members individually.
list_member_hook<> is very troublesome in MSVC and is known to cause problems in other compilers when used inside a class which has a virtual base class.
This fixes an issue with adding/removing tracks while auditioning.
Session::remove_routes() calls Graph::clear_other_chain(),
which will block until the graph chains have been swapped.
Capture latency needs to be updated before playback latency,
various internal parts depend on this order (which is also
the default for libjack itself).
Retain min/max when copying inherited latency from
connected ports.
When there is a direct connection port A out -> port B in,
min/max latency range should be retained in direction
of signal flow.
Ardour can only adjust latency in reverse direction of signal flow
to match min = max with an internal delay-line. ie set playback
latency of an input port, or capture latency of an output port.
This allows to get public latency ranges for any port
including Ardour's latency-compensation as well as including
any latency induced by external JACK applications.
It is mainly useful to detect ambiguous latency and to
forward public port latency of connected ports.
When switching backends, the effective sample-rate is zero.
This only affects the butler thread (the only active thread when
stopped). The actual issue here is the butler calling
"non-realtime-stop" without a backend. However fixing 0/0
generally seems appropriate.
```
#0 in int_div_round<long>(long, long) (x=0, y=0) at ../libs/pbd/pbd/integer_division.h:36
#1 in Temporal::samples_to_superclock(int64_t, int) (samples=0, sr=0) at ../libs/temporal/temporal/superclock.h:39
#2 in Temporal::timepos_t::timepos_t(long) (this=0x7f94bc0a5890, s=0) at ../libs/temporal/temporal/timeline.h:55
#3 in ARDOUR::Automatable::non_realtime_locate(long) (this=0x55a12a980cc8, now=0) at ../libs/ardour/automatable.cc:421
#4 in ARDOUR::Route::non_realtime_locate(long) (this=0x55a12a980ae0, pos=0) at ../libs/ardour/route.cc:5462
#5 in ARDOUR::Session::non_realtime_stop(bool, int, bool&) (this=0x55a12e0cd000, abort=false, on_entry=1, finished=@0x7f94bc0a5e0f: true) at ../libs/ardour/session_transport.cc:1487
#6 in ARDOUR::Session::butler_transport_work(bool) (this=0x55a12e0cd000, have_process_lock=false) at ../libs/ardour/session_transport.cc:1153
#7 in ARDOUR::Butler::thread_work() (this=0x55a12f3b7000) at ../libs/ardour/butler.cc:222
#8 in ARDOUR::Butler::_thread_work(void*) (arg=0x55a12f3b7000) at ../libs/ardour/butler.cc:16
```
There are cases where gtk modules log critical messages,
notably gtk-2.0/modules/libgail.so can trigger
`g_log: g_object_unref: assertion 'G_IS_OBJECT (object)' failed`,
which should not cause Ardour to terminate.
Somewhat unbelievably, macOS defines size_t as unsigned long, which the compiler there
believes is distinct from unsigned long long even then they have the same bitwidth
and signedness. We don't have any string converters for unsigned long, only uint64_t
(unsigned long long), so various things break.
Firstly, ::compute_bounding_box() cannot allocate space, so do not try to do that here.
Secondly, if there is no allocation yet, use the CairoWidget's own size_request()
