When loading the state of a Route fails (here
"unknown enumerator SurroundMaster in PBD::EnumWriter"), the
routes which have already been loaded are not added to the
Session's routelist.
Already existing routes that have an InternalSend or have
a circular reference:
The Send's `_send_from` holds a shared pointer
`<Route>(shared_from_this())` to the Route, and the
Route's ProcessorList contains the InternalSend.
This leads to various
"SessionHandleRef exists across session deletion" of
IO, Ports, GraphNode, Graph, etc
which causes issues when loading another session.
Session::destroy() cleans calls drop_references for
each route in the RouteList, which breaks the circular
dependency (InternalSend drops reference to Route).
But here the RouteList is empty.
Crash fixed:
* Load a session that fails to load a Route
(here a session created on with the vapor branch, on master)
* Then load another session without restarting Ardour.
When copying a mono LV2 plugin to a stereo track, the
state of the copied mono plugin is copied to the replicated
instance.
However at this point in time PBD::Stateful::ForceIDRegeneration
is still enabled, and the state ID has not been set.
This used to not be an issue. Older versions of liblilv
handle non-existent paths just fine.
However in lilv v0.24.20-10-gdd5e851 `lilv_path_absolute`
was replaced with `zix_canonical_path` which returns NULL
if a state file does not [yet] exist. This lead to a segfault
due to strlen(NULL) in `serd_node_new_file_uri`:
#0 strlen -- SEGV on unknown address 0x000000000000
#1 serd_node_new_file_uri () at /lib/x86_64-linux-gnu/libserd-0.so.0
#2 lilv_state_new_from_file () at /lib/x86_64-linux-gnu/liblilv-0.so.0
#3 ARDOUR::LV2Plugin::set_state(XMLNode const&, int) at ../libs/ardour/lv2_plugin.cc:2320
* yabridge runs the plugin's process function in a dedicated
bridged thread. Ardour's process thread is not (it just waits)
* When a plugin calls `restartComponent` from the realtime
thread. yabridge uses a host notification thread to perform
the callback.
Unlike other VST3 implementations that use a notification thread
(eg. JUCE), yabridge blocks and waits for the notification to
complete before the realtime thread can continue.
This leads to a deadlock.
However, we know that yabridge always synchronizes the
callback and concurrent calls are prevented by yabridge's design.
https://github.com/robbert-vdh/yabridge/issues/266
Locations::ripple can never add/remove markers, hence
Locations::changed is not applicable.
That signal is to indicate when more than one location is
added or removed from the location list.
Undo sets the state of ALL Locations, which resulted in
at least two signals for each Location (name changed,
start+end changed), even if there was no change.
This is in preparation to reduce signals during
Location Drag motion (which operates on a copied
Location, that causes [static] signal emissions,
but the location itself is not in any list).
Every call to ::next_section() copies the location list
and sorts all the regions.
If the session has a significant amount of Locations and
Section Marker (#9568 has 300+) sorting them each time
when iterating over sections is significant.
Exporting a dedicated Range should not be constrained by
session markers. This fixes at least 2 issues:
* export could be cut short when using latent plugins
* exporting a range crossing the session end marker was cut short
When a user first connected a port to the Timecode master
input and then disconnected a previous port, the Timecode
master assumed it was not connected.
Previously the port-engine was a LIFO. Changes were pushed back
and then popped-back. This causes issues when re-connecting
Transport Masters.
The GUI does the following when changing connections:
1. disconnect all
2. connect to new port
which lead to TransportMaster::connection_handler being called
in reverse order: connect, disconnect, and the transport
master was assumed to not be connected.
--
Now connections queue is a FIFO and code was consolidated.
(Note, we cannot use a std::deque because it does not support
memory pre-allocation with ::reserve)
The former was incorrectly implemented, and the latter has already been tested more
in real life.
We should likely remove ::remove_time also and use shift() there too, but that
requires testing negative shifts more broadly.
Ardour accidentally removed nascent source-files during cleanup.
This can lead to missing files when recording directly after a
cleanup.
This also ensures that there are no duplicates in the
dead-sources (file sources unused in the current snapshot) list.
parens were in the wrong place - we need to add the ::magnitude() of
the tick-based duration AFTER conversion of audio-time position to beats, not
before.
we control the CC number sent by launchpad faders, and 0x20 is too large
because it causes the faders to overlap with some of the CC values
sent by RHS pads. Parametize the first fader and use it everywhere
This was causing issues with cues when clips were set to gate triggering. A pad
long press was causing the clip to release. Additionally, remove an extra timer
that was being set.
This is not complete, because the symol names are identical, and there's no way (yet)
to ensure which versions Ardour will use if both are dynamically loaded.
This can happen with snapshots or after save-as with
.ardour session files having different "Names" in the same path.
Or simply by saving a session on macOS in /tmp (which is really
/private/tmp).
This likely needs checking for all surfaces that inherit from MidiSurface. It is clearly
the correct thing to have in the code, but existing behavior might be predicated on
the former incorrect connection
This is perhaps a better solution than b8551eed7e
and 8d0a655608 and 7942897d93. It is certainly less
fragile.
It is more consistent with other plugin standards,
where modules are closed with the last instance in a session.
Then again keeping the VST3 factory around is beneficial
when switching snapshots.
Discuss, and let's watch for issues when re-loading a
previously unloaded VST3 module.
There is no thread when an AbstractUI<T> is constructed. The event loop name and the
association between the event loop object and the thread that "runs" it must be
set from within the thread, which is not created until BaseUI::run() is called.
There appears to have been some confusion in e3569b64 about how this
all works; this commit should remove that
e.g. selecting a track causes a ControlNotFoundException
if the ctrl surface is enabled, but no hardware is connected.
terminate called after throwing an instance of 'ArdourSurface::ControlNotFoundExceptio
```
#0 0x00007ffff14d8c2e in __cxa_throw () at /lib/x86_64-linux-gnu/libstdc++.so.6
#1 0x00007fffe2b560a0 in ArdourSurface::Console1::get_button(ArdourSurface::Console1::ControllerID) const (this=0x61d0017e1c80, id=ArdourSurface::Console1::FOCUS1)
at ../libs/surfaces/console1/console1.cc:928
#2 0x00007fffe2bfc647 in ArdourSurface::Console1::map_select() (this=0x61d0017e1c80) at ../libs/surfaces/console1/c1_operations.cc:653
#3 0x00007fffe2b55384 in ArdourSurface::Console1::map_stripable_state() (this=0x61d0017e1c80) at ../libs/surfaces/console1/console1.cc:832
#4 0x00007fffe2b541ab in ArdourSurface::Console1::set_current_stripable(std::shared_ptr<ARDOUR::Stripable>)
...
```
This does at least fix the crash. Ideally the surface would
only be enabled if there is hardware present.
Some plugins call restartComponent(Vst::kParamTitlesChanged)
when their GUI is created, from the call that creates the UI.
This lead to a stack-overflow recursion in Ardour:
ProcessorBox::redisplay_processors -> VST3Plugin::has_editor
-> [plugin] -> VST3::restartComponent -> signal proc changed
-> ProcessorBox::redisplay processors
Since 62fc1d3c2e, Delivery buffers were flushed twice.
Once by copy_to_outputs() and again later by
Delivery::flush_buffers. This resulted in duplicate events
during export (see 576840c09e, MIDI buffers are not cleared
after flush to allow export processing to grab the data from
the port-buffers).
The workaround in 62fc1d3c2e is only relevant for ClickIO,
other Deliveries (Send is a Delivery) are explicitly flushed
by Route::flush_processor_buffers_locked.
This reverts commit 615326be9b because it
breaks windows builds.
```
File "/home/ardour/ardour-w64/wscript", line 1462, in configure
set_compiler_flags (conf, Options.options)
File "/home/ardour/ardour-w64/wscript", line 522, in set_compiler_flags
if re.search ('x86_64-w64', conf.env['CC']) is not None:
File "/usr/lib/python2.7/re.py", line 146, in search
return _compile(pattern, flags).search(string)
TypeError: expected string or buffer
```
Region::fade_range emits a signal which will call
ARDOUR::Playlist::region_bounds_changed, which takes a WriteLock.
That resulted in a WriteLock with ReadLock held.
CC is already set to a string. (And if it ever should be None, we want
to handle that explicitly.)
(And #autowaf.display_msg handle Booleans just fine.)
The sub_config_and_use function recursed, but it also invoked
autowaf.set_local_lib , which however didn't do anything useful. The
HAVE_ defines are not used anywhere, and the AUTOWAF_LOCAL defines are
only used in autowaf.use_lib, which however isn't used anywhere.
Dropping these defines simplify the build environment and makes the
compiler command line half as long and thus makes debugging much more
manageable.
https://waf.io/book/ says
By default, the project name and version are set to noname and 1.0. To
change them, it is necessary to provide two additional variables in
the top-level project file
- and waf code inspection confirms that waf itself only will use the top
level APPNAME.
autowaf has no real shutdown functionality anyway. The automatic
shutdown function that could have been called wouldn't work anyway, as
it takes an argument.
The only reason it doesn't fail is that the top level wscript has no
shutdown handling and doesn't recurse to other scripts, so it is all
dead code.
It turns out that slightly older versions of ALSA create different "pretty"
port names for USB MIDI devices than slightly newer ones. The new versions
use names that match those seen on other platforms.
This means that to do port matching on Linux now requires a regexp
to match the possible alternatives. This matters much more for the LPP,
which has 3 input ports and 3 output ports, than it does for most devices
that have a single input and single output, and we can "find" the ports
just using simple string searching
The kissfft library changed its SONAME and how the various libraries
were organized, e.g. the functions for complex and real numbers aren’t
in separate libraries anymore.
Signed-off-by: Nils Philippsen <nils@tiptoe.de>
when Session emits RouteAdded, each handler (editor, mixer, trigger pages etc) will
execute their callbacks in order. But Editor::add_routes() selects the routes too,
which triggers a PresentationInfo::Change signal. This is received by e.g. the Mixer
before it has even found out about the newly added stripables. This in turn
leads to severe confusion regarding the state of the selection in the mixer,
and potentially elsewhere.
So, just add a PresentationInfo::ChangeSuspender for the scope of the RouteAdde
signal emission
Initially thew new region has a length of zero (0:0). When
merging Notes from a truncated region (no explicit note-off)
those notes are lost:
"Stuck note resolution - end time @ 0:0 is before note on: @ 0:0"
Truncate (or split) a region so that a note is cut short:
```
Region [{<--Note-------->}]
```
becomes
```
Region [{<--Note--]
^ implicit note-off at region boundary
```
When combining this region with an empty Region after gap,
```
Region [{<--Note--] [ ]
```
the result should be
```
Region [{<--Note->} ]
```
For this reason, even without a gap between the regions,
the original note length must not be restored.
The result MUST NOT be the same as the original:
```
Region [{<--Note-------->}]
```
Playlist::split can result in removal of a region
(adding two others instead). In this case the state
of the removed region (if modified) is not saved.
I/O plugin Controls are destroyed with ~IOPlug, however
Session::destroy()'s call to drop_references() still
triggers AutomationControl::session_going_away() on the
binding proxy.
This is even properly documented in session_object.h:
> A named object associated with a Session. Objects derived
> from this class are expected to be destroyed before the
> session calls drop_references().
Playlist::add_region, Playlist::add_region_internal always
adds the region to the top of the playlist, ignoring the
region's layer.
Note that there is also difference between
Region::layer and Region::layering_index.
Previously creating a new preset first attempted to load
it from disk, before creating it. This resulted in a
`XML error: failed to load external entity`.
The definitions from <stdatomic.h> conflict with <atomic>, which causes
a build failure since <atomic> is included previously.
Signed-off-by: Violet Purcell <vimproved@inventati.org>
This one is more complex than the Beats or superclock variants, because
we cannot just start from the front of the map. Instead, we have to
first iterate through the map so that we start the code in
_get_tempo_and_meter<...> from the TempoPoint and MeterPoint
in effect at the BBT_Argument's reference time.
Track::use_captured_midi_sources is called from use_captured_sources().
which is only called from DiskWriter::transport_stopped_wallclock(),
which is only called from Session::non_realtime_stop().
This solves several issues related to splitting or pasting regions, when
there is more than one layer.
Rather than assign a new group-id for "all the regions on the right of a
split", only ions that had a *prior* group-relationship should be
propagated into the new group.
Signed-off-by: Robin Gareus <robin@gareus.org>
LLVM libc++ does not have the ext/atomicity.h header. This fix is copied
from the upstream vst3_pluginterfaces repo.
Signed-off-by: Violet Purcell <vimproved@inventati.org>
The alias was only used when it was exposed in lua. It was without any
indication that it was a deprecated alias, but let's just bite the
bullet and get rid of it.
A group of functionality was only used once, in
TempoPoint::quarters_at_superclock . Keep things simple and enable
further refactoring and cleanup by inlining everything and dropping
superbeat, big_numerator and super_note_type_per_second from Tempo.
The use of big_numerator right next to superclock_ticks_per_second
seems error prone. It should perhaps just be refactored to work in
superclock domain all the time.
It seems weird that the ramped case is much simpler than the non-ramped.
This (pretty much) removes the last references to "superbeat", which
I thus doesn't have to understand ;-)
The note_type_as_beats was the only temporal thing using hardcoded value
of 1920. It seems like it just should use the usual Ardour PPQN (aka
ticks_per_beat) ... which also has the value 1920.
It is however not used after d77db816de.
There is no need for scts_set now. "Early" use of SCTS will just give the
value 0. DEBUG_EARLY_SCTS_USE can thus just check that
_superclock_ticks_per_second doesn't have the initial value of 0.
If DEBUG_EARLY_SCTS_USE somehow was set, compilation would fail because
of includes inside a namespace.
(Even without DEBUG_EARLY_SCTS_USE, any early use of superclock will
probably fail clearly with division by zero. There is thus not much need
for DEBUG_EARLY_SCTS_USE now.)
Make sure all code paths that use Temporal will initialize and reset it
properly. Some code paths (in tet runners) doesn't use Sessions, so
Temporal::reset() has to be invoked directly.
Just set the static superclock variable to 0 as initial value.
TempoMap will still be initialized early as a singleton, but we
introduce a new constructor so it is created empty (and thus not really
usable until Temporal::reset() or similar has populated it).
We can thus drop the static initialization of superclock. The default
superclock rate of 282240000 will now only live in Temporal::reset().
With this change there should no longer be any uninitialized use of
superclock_ticks_per_second(), and there should not be any problems for
DEBUG_EARLY_SCTS_USE to catch. (It is however broken in other ways -
that will be fixed next.)
The superclock rate is variable and is stored in session files since
7.0 . It is set globally when reading from a session file, and it really
should be reset to a known state whenever a new blank session is
created from an existing one. Currently, that doesn't happen. We fix
that by resetting it in Temporal::reset().
For now, we are duplicating the default superclock rate 282240000 from
superclock.h , but we can drop that when all code paths that use
superclock also use Temporal::reset().
This will provide an (extra) guarantee that
set_superclock_ticks_per_second() always is invoked before creating
TempoMap or using superclock_ticks_per_second() in other ways. The
DEBUG_EARLY_SCTS_USE in superclock.h is thus closer to passing.
Ardour uses some global variables and singletons. These global variables
can be initialized with a value prior to program execution (especially
if they are const), but some of the static variables are modified, and
it is crucial that they always are reset when switching to another
session. To keep things simple and explicit and consistent, we thus
introduce Temporal::reset() to reset TempoMap (and later on also the
superclock rate). This is somewhat similar to Temporal::init(), which
usually only is invoked once (on program start) to initialize singletons
(such as the TempoMap).
9964f20c added TempoMap initialization to Session::create() ... but only
when not using a template. This create method is mainly preparing the
filesystem for a new session, and TempoMap initialization doesn't seem
like a perfect fit for it. It also seemed odd that it only initialized
TempoMap for clean new sessions, while existing sessions and templates
initialized it elsewhere.
Instead, invoke the TempoMap initialization early in the Session
creation process. This might introduce an extra and unnecessary TempoMap
initialization when loading an existing session or using a template, but
that will be cheap and do no harm, while providing a guarantee that we
always use the same default value.
Robin Gareus
Ben Loftis
Edgar Aichinger
John Emmas
Paul Davis
Nils Philippsen
dominik martinez
Martin Vlk
Holger Dehnhardt
Mads Kiilerich
Krzysztof Gajdemski
Alexandre Prokoudine
JungHee Lee
Violet Purcell