Just an update to slightly rotten wscripts, shouldn't be any changes during an
ardour build. Motivation being a short development cycle for working on evoral
and/or its test suite.
Note the old Note::operator= was unsafe, since it made shallow copies of the on
and off events, which results in a double delete of events when the notes are
destructed.
I'm not sure if this is really the best way to do event types (should it
just be a completely static enum in evoral, or completely dynamic and
provided by the type map, or a mix like currently?), but previously the
event type was frequently set to either total garbage, or parameter
types, which are a different thing.
This fixes all those cases, and makes Evoral::EventType an enum so the
compiler will warn about implicit conversions from int.
It is slightly questionable whether type specific methods like
velocity() belong on Event at all, these may be better off as free
functions. However the code currently uses them as methods in many
places, and it seems like a step in the right direction, since, for
example, we might some day have events that have a velocity but aren't
stored as MIDI messages (e.g. if Ardour uses an internal musical model
that is more expressive).
In any case, the former inheritance and plethora of sloppy casts is
definitely not the right thing.
eg. import a .mid that has a CC later in the file.
Arodur wrongly added an initial point, effectively moving the event
backwards to "0" (no virgin territory)
uselib is no longer implicit (inherited by .use). This is still incomplete,
some uselibs for non-linux variants may be missing.
bld.is_defined("HAVE_XXX") also no longer works and will have to be
changed (I think to bld.env["HAVE_XXX"]) in countless places.
libevoral itself doesn't seem to need libpthread - but by some mechanism it #includes <pbd/event_loop.h> (which now does #include <pthread.h>). So let's make sure it can be found.
- disallow simultaneous events via ControlList::editor_add ()
- clicking on an automation line selects the points that define it.
- don't 'flash' a region selection when using mousedraw mode.
- cp click selection resembles region selection.
- region gain points respect snap modifier (a la automation points).
PluginInsert::automation_run() subdivides plugin-run on every
control-port automation event (without splitting the process cycle).
libevoral has no automation-control context, hence this function
must be implemented by Automatable.
Iterating over a const Midi-Sequence calls Evoral::Sequence::set_event(),
which in turn used Evoral::Event::operator=() which always created
a new event-ID (create copy of the event).
Issues fixed:
- Saving *unmodified* MIDI produced new event-IDs on every save;
files changed with every save. - greetings to Deva.
- all [GUI] operations that use IDs to refer to notes e.g. undo.
invalid undo-history.
Also clarify assignment operator name. Prefer explicit assign() over =.
Evoral::Beats::operator>() rounds to (1.0 / PPQN), hardcoded 1/1920.0.
If the time difference between two events is smaller than 1/PPQN,
Beats::operator>() and Beats::operator<() produce ambiguous results.
The same pair of values is both "less than" and "greater than" depending
which operator is used.
While it's fine for some cases to ignore the order of nearly concurent
events, the std::priority_queue must be strictly ordered.
- don't keep setting/unsetting write pass when transport frame
remains the same (think larger jack buffer sizes)
- insert guards are now 64 frames after when.
- refactor previous approach.
- clearing automation points sets control to "off" rather than touch.
- multiple touches on the same pass acts consistently (no more
fader jumps on mouse button press
- use actual value for initial point rather than some arbitrary
default. clarify new semantics of add () (with_default->with_initial).
- clean some whitespace
- add guard points as needed in stop.
- catch grab broken signal (i can't trigger it, but the docs seem
to think it is essential).
- probably fixes a lot of cases where note ids are assumed to be
unique (they weren't for copies and some others).
- wrong branch, but it needs testing.
towards fixing #6238 and #6096.
GUI thread:
#2 Glib::Threads::Mutex::Lock::Lock
#3 Evoral::ControlList::eval
#4 Evoral::Control::get_double
#5 ARDOUR::AutomationControl::get_value
#6 ProcessorEntry::Control::control_changed
..
#15 PBD::Timer::timeout_handler
at the same time: Audio Thread (try-lock, fails)
#0 Evoral::Curve::rt_safe_get_vector
#1 ARDOUR::Amp::setup_gain_automation
#2 ARDOUR::Route::process_output_buffers
Due to the failed try-lock.. AMP::_apply_gain_automation
is false. and Amp::run() uses a different gain factor.
-> click.
Fixes bug #6166 (except record).
This attempts to follow the "current" control value somewhat aggressively:
* On locate, slider is set to the value from the top region at the new
transport position.
* Playback or MIDI input is followed "live".
* Whenever the slider is moved (including automatically), that value is emitted
as an immediate event to keep external gear in sync.
General idea is that the Ardour slider should act as a mirror of an external
hardware knob, and both should be synced to whatever the control is at the
current transport position. Since we lack real playback/touch/etc modes for
these for now, we must choose one behaviour, and this seems like the most
reasonable one.
Follow is handled in the audio thread, which is probably not ideal, but since
these controls have no lists and do not record, should be fine. Probably.
This avoids stuck notes if active notes are edited, but without stopping all
active notes in the region on any edit as before.
This implementation injects note ons in places that aren't actually note
starts. Depending on how percussive the instrument is, this may not be
desired. In the future, an option for this would be an improvement, but there
are other places where "start notes in the middle" is a reasonable option. I
think that should be handled universally if we're to do it at all, so not
considering it a part of this fix for now.
Best to just do this as early as possible to avoid having to deal with this
situation all over the code.
Also fixes violation of LV2 MIDI specification, which requires no such events
are delivered to plugins.
Silly to make a junk Note just to pass to append_note_off_unlocked, which just
uses the fields that are on the MIDIEvent anyway then throws it away.
Also explicitly dispatch to append_note_off_unlocked in the caller for note ons
with velocity 0 rather than make append_note_on_unlocked deal with it.
Add a test, based on the worked example in www.korf.co.uk/spline.pdf, for
the constrained cubic spline interpolation.
The delta values for the float comparisons are rather arbitrary, I'm sorry
to say: they're basically chosen so that everything passes.
When the crossfade length is only 1 frame, I got strange
gain coefficients from get_vector (63 in my case).
The function wrongly returned the x axis value.
We're still a very long way from tolerant of weird SMF files (libsmf takes a
"crash if input is not exactly perfect" philosophy, if we're going to be polite
and elevate such a thing to "philosophy"), but at least we'll get what's there
from files truncated by old broken versions of Ardour or other situations.
This was a very clever attempt to fix a non-problem. If the platform doesn't have enough file descriptors available
then the platform is broken and we're not going to hack around trying to fix it.
For things like copying from pitch bender to a CC.
Also things like fader to pan, but that seems a bit funny. The conversion
probably needs to be a bit smarter here, perhaps taking the normal into
consideration...
Among other things, this means that automation controls/lists have the actual
min/max/normal/toggled of parameters, and not those inferred from the Parameter
ID, which is not correct for things like plugin parameters.
Pushing things down to the Evoral::ParmeterDescriptor may be useful in the
future to have lists do smarter things based on parameter range, but currently
I have just pushed down the above-mentioned currently used attributes.
Remove old (already #if 0'ed) implementation of Evoral::coverage() and its
comments.
Tidy up the comment enumerating all the possible ways in which two ranges
can overlap, note the Evoral::OverlapType corresponding to each one, and add
comments to the if()s in coverage corresponding to the cases in the list of
overlap types.
Remove some commented-out assert()s that actually do happen, and re-instate
one that really shouldn't.
Fix a small typo (with -> within)
The various conditionals in Evoral::RangeList::subtract() appear to have
been there to work around
(a) coverage() not always returning the correct value, &
(b) the test suite assuming that the ->to point lies outside the range
Now that these are both fixed, the implementation of subtract() becomes
quite a bit clearer. I replaced the if()s with assert()s for now, but these
shouldn't trip if coverage() is working as I expect.
Also (attempt to) clarify the comments in subtract.
Rewrite Evoral::coverage() to (hopefully) do what it's supposed to.
Return OverlapNone for invalid ranges: if either of the ranges passed to
Evoral::coverage() have negative length (i.e. start > end), return OverlapNone
- it seems reasonable to say that a negative-length range can't overlap
anything. Also return OverlapNone from the fallthrough case, though this should
never happen.
Some of the tests for Evoral::RangeList::subtract() assume that ranges
don't contain their end (->to) point. This appears inconsistent with how
they are used elsewhere.
Add some ASCII art comments to the tests to try to clarify what they're
really testing for, and amend subtractTest1, subtractTest4, & subtractTest5
to incorporate the assumption that ranges include their end points.
This is not used anywhere in Evoral and is just a wrapper around the PBD
RingBuffer anyway. Towards a (once again?) independently buildable/testable
Evoral and fewer cross-dependencies.
Add a test function to test Evoral::coverage() with all possible overlap
types. The first test (line 161) that expects OverlapExternal will fail
with the current implementation of coverage().
There's possibly still a discussion to be had about what the overlap type of
ranges with negative lengths should be: there are currently places in the main
Ardour code base where coverage() is called with ranges where start > end.
Fix compile errors in libs/evoral/test/, by explicitly calling
Evoral::MusicalTime::to_double() wherever a double value is required of a
MusicalTime.
Some of the double variables should probably really be made into MusicalTime
ones instead, but I don't want to mess with this too much.
takeFiveTest still fails for me after this, but a failing test is probably
more informative in the long run than a test that won't even compile.
This lets us get a more explicit handle on time conversions, and is the main
step towards using actual beat:tick time and getting away from floating point
precision problems.
Fix initial read of discrete MIDI controllers.
Fix spurious note offs when starting to play in the middle of a note.
Faster search for initial event when cached iterator is invalid.
So much for dropping the cached iterator. The iterator is responsible for
handling note offs, so that doesn't work. This design means we have some stuck
note issues at the source read level, but they should be taken care of by the
state tracker anyway.
This should probably hijack the same modifier as the guard points and work the
same on all automation tracks, but I did it this way to not change behaviour of
track automation where a default is much more reasonable.
This cleans up a lot of false-positives in static analysis
and also helps compilers to optimize code paths in general.
(tagging the fatal stingstream operator as ‘noreturn’ is
far less trivial)
Serialization of Session::save_state() will already protect against most of this, but there is really no
good reason why Evoral::SMF's API should require single-threaded/explicit serialization.
Also some work towards tolerating automation controls with no automation list,
towards actually doing something for these cases, though not required just to
fix this crash (MidiTrack::set_parameter_automation_state() avoids those
paths).
It'd be nice if we could use 'ARDOUR::config_dir_name' for this purpose (or perhaps 'PROGRAM_VERSION'). However, neither is implemented widely enough at present to make this practical. Keep an eye on them though, as possible future strategies.
TODO: needs undo. only works in top quarter of automation lane. selection model feels weird sometimes. needs to show gain curve when you are using Range tool
Constrain control points to one per tick (1/1920 beats).
Prior to this it was possible to set two values to the
same time (interpolation and iteration failed).
