Custom changes made to the config after first initialization got lost
after setting work directory. Therefore moved custom initialization into
a separate function which is called after setting work directory.
The local binary directory should be more prioritized regarding
file inclusions. Otherwise for example the global embedded_resources.h
gets included instead of the plugin-specific one.
Printing error messages using fprintf() does not help much with
RemotePlugins as their stdout/stderr usually is not redirected to the
parent process. Instead use the debugMessage() function.
On non-Linux platforms configuration file was saved in the application
path which definitely is not desired and also will not work in many
situations. Therefore always save ZynAddSubFX configuration file in
LMMS working directory.
Added a window icon to ZynAddSubFX windows for better appearance.
This is currently only possible for the win32 version a lot more hacks
are required to achieve this for X as well :-(
It's now possible to hide the ZynAddSubFX UI by simply closing its main
window instead of going back to LMMS and toggle the "Show UI" button.
Furthermore moved code for GUI thread into a non-static member function
of RemoteZynAddSubFx and removed the old code for handling IdHideUI
messages.
We must not have any file called config.h in directories which are added
to global include directories. For example when building the CALF plugin
the file config.h from libsamplerate was used rather than CALF's local
one. Fixes missing LADSPA symbols in CALF library if no usable
libsamplerate was found in the system at configure time.
Also bumped version number in src_config.h.
When calling TrackContentObject::paste(), only the content should be
pasted, not the position. This fixes pasting for example note patterns
and is also a more generic solution for the bug fixed in commit
4ec5c6afe1.
Closes#3005649.
Not sure when it broke but cloning BB tracks erroneously cloned
positions of TCOs as well which made the source BB track have
two TCOs one upon the other while the destination BB did not have
any TCOs at all.
The fix is to save the position of the destination TCO before and
restore it manually after paste().
Commit 884b9ca671 completely broke the
note detuning feature as the detuning value was not incoorporated into
the frequency calculation anymore. Fix this by adding the detuning value
to the pitch variable.
* calf-updates:
Fix incorrect plugin class (spotted by Luis Garrido).
Add windowing-based antialiasing to hard sync in Monosynth; update last stretch even if synth is idle. (cherry picked from commit d64a0487d88f6730ef90dd6d79b892ad4a114138)
The compiler flag -ftree-vectorize brings only little advantages in
some special cases. In all other cases either worse or even wrong code is
generated due to bugs in GCC (especially for Win32). Therefore remove it.
(cherry picked from commit 47f56ed460)
Various plugins were missing the PLUGIN_EXPORT declaration with their
PluginDescriptor instantiations and lmms_plugin_main implementation
which made some plugins not show up with latest Win32 build.
(cherry picked from commit f85898ba6e)
Silence a compiler warning introduced by custom changes and fix a
possible compile failure by including another header file in a header
file that we cleared before.
As the colors of QToolTip depend on the desktop environment used,
explicitely set background and text color for fixing cases like
white text on bright yellow background.
Closes#3006955.
Integrates Sami Boukortt's patch described below:
I have made a patch for the “Save changes?” dialog box to let the user choose
between “Save”, “Discard” and “Cancel” instead of “Yes”, “No” and “Cancel”.
I find this way more intuitive, since it enables Qt to reorganize the buttons
properly according to the platform and to give them an icon (at least on X11).
It also allows users to know what is going to happen without having to read
the question.
Now that we're not removing the config.h-inclusions from upstream's
source files we have to provide a dummy config.h. At the same time this
allows us to move macro definitions from CMakeLists.txt to config.h.
There have been massive changes in CALF's master branch. It makes no
sense to manually cherry-pick all these commits. Therefore rebased the
CALF codebase in LMMS to the one found in CALF master branch.
Thanks to Fryderyk Dziarmagowski for providing a patch for the .desktop
file of LMMS. It should be standard conform now. An icon is now
explicitely installed in /usr/share/pixmaps so the .desktop and menu
files can specify a generic icon name rather than an absolute file path
to a PNG file of LMMS' default theme.
It's a bad style to manage global instances of an object in the object
itself. Therefore introduced a nested helper class which manages all
instances of EnvelopeAndLfoParameters globally. It is now responsible
for global reset/triggers of LFOs. In contrast to previous state, this
is now done thread-safe. Fixes crashes for example while importing
MIDI files.
The new deleteHelper() template function takes a pointer, saves it
to a temporary, sets the passed pointer to NULL and then deletes the
object it was referring to before. This way we can spot bugs caused by
undesired references to global objects at shutdown more easily.
It was not safe to immediately delete the AutomationPattern attached to
an InlineAutomation object before commit "InlineAutomation: delete
attached AutomationPattern immediately" as the AutomationPattern
destructor sometimes fiddles around with AutomationEditor (GUI!).
However this has changed now and this enables us to delete the
AutomationPattern immediately. This fixes some random crashes with
scheduling AutomationPatterns for later deletion.
Closes#2982696.
It was not possible to call any of the functions of AutomationEditor
from threads other than the main GUI thread as the m_pattern pointer
was not protected by a mutex. It is however desired to be able to call
for example AutomationEditor::setCurrentPattern(...) from different
threads (mainly required when deleting AutomationPatterns in other
threads). Thus made this method and all accesses to the m_pattern
pointer thread-safe.
The global channel link checkbox did not work as expected as of the
M/V-split during 0.3.x -> 0.4.x development. Connected the slot
responsible for updating individual ports after toggling checkbox and
fixed the logic behind.
Closes#2964800.
