Whenever a variable gets declared using 'thread_local' MSVC requires that it should not be compiled with DLL linkage (i.e. it mustn't be exportable). So for Temporal::TempoMap we'll need to export the required members individually, rather than exporting the entire class.
Now that we use superclock_t for audio time, it is possible for the square of an audio time value to overflow int64_t quite easily.
This change fixes that (and cleans up other code a little), but probably a different solution would be a good idea
Later I'll need to push some extra changes (to support 'tempo_map_p' and 'boost::intrusive::list' etc) but these initial ones (hopefully!) won't cause any issues for the other builds.
These files will be removed at session close if they are still empty. Their existence prevents
collisions across snapshots and likely some other wierdness that relied on file existence for
various tests and conditions
resort_routes() is a no-op during session destruction,
however _current_route_graph can still hold references
to routes. A GraphVertex is-a boost::shared_ptr<Route>.
The i18n header needs to be included last, since it defines
`_` which results in conflicts with other headers.
see #8361 and e.g.
boost/function_types/detail/class_transform.hpp:23:26:
error: ‘boost::mpl::placeholders::_’ has not been declared
using mpl::placeholders::_;
When a track's output is not connected, its alignment
can /float/. Previously it assumed the worst (compensate
for worst-case latency).
The track itself has no playback latency, so any send would
delay the 'thru' path to match the send's playback latency.
This resulted in the track reporting a non-zero latency which
is not really present, nor relevant.
With glib 2.64, volatile gint produce annoying Wcast-qual
compiler warnings with mingw/gcc-8.
Even though the the build succeeds, and only glib 2.68+
requires non-volatile atomics this results in a much
cleaner build-log.
