The `MACRO(...)` macro stores the macro in progmem, and as such, needs to use
constants only. Rework the logic in `macroAppSwitch()` so that we keep that
requirement.
For some reason, the previous version of the code worked with gcc 5.4, but not
with gcc 7+, and the latter is right to complain.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Sketch exploration choked on the rare case that an empty keymap
was supplied with `KEYMAP()`.
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>
To be able to compile a firmware for the Raise, we need to include the HID
facade, and the base keyscanner implementation from the Raise device plugin.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
This was done to enable separate inclusion of the central runtime
class without having to include the central header
kaleidoscope/Kaleidoscope.h which used to pull in a lot of stuff that is
not required in many compilation units.
The new class `Runtime_` lives in namespace kaleidoscope its singleton
instance is `kaleidoscope::Runtime`. It is now only available internally
in library Kaleidoscope but not from the sketch.
The original class name `Kaleidoscope_` in global scope has been deprecated.
The original instance name `Kaleidoscope` in global scope has been
preserved to be used by end users in their sketches.
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>
This implements an extension to the `TapDance` plugin, allowing us to store
`tapDanceActionKeys()`-esque lists in Storage. The core idea here is very
similar to that of `DynamicMacros`: we dump/restore the full list via Focus,
build a cached index on setup and any updates, and play back the selected key
when need be.
Unlike `DynamicMacros`,this plugin is built on top of `TapDance` and cannot
function without it.
Fixes#730.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
The Makefile in Model01-Firmware was adapted quite a while
ago in a way that it simplifies builds in hardware paths that are not
below Arduino's standard directories.
This simply copies the makefile from the Model01-Firmware repo.
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>
The virtual hid was recently included in the core firmware repo.
Due to missing `#ifdef KALEIDOSCOPE_VIRTUAL_BUILD` in some
files, two versions of the hid library, one for the virtual and one for
the physical device were build. The linker perferred the one that it
encountered first. That caused a link order dependency that possibly
renders some firmware builds without HID reports being send to the host.
This change adds thse missing `#ifdef KALEIDOSCOPE_VIRTUAL_BUILD`
clauses.
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>
This static member did not have an instance under
certain circumstances. A typical workaround for such
a situation is to make it a local static of an accessor
function. By this means the one definition rule is
not violated and the object always instanciated. It is still optimized
away by the compiler in case device::Base<...>::LEDs() is not called.
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>
`RaiseKeyScanner::actOnMatrixScan()` erroneously looped through all the columns
in the matrix, while we only wanted to loop through the left half, since we read
the right half at the same time, not separately. This resulted in half the keys
producing two events, due to overflow. We also calculated the key number
incorrectly.
Both of these issues are fixed with this patch.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
The submember is a bitfield that needs its own braced list.
Gcc silently tolerates this clang warns about it.
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>
Clang seems to have problem with recognizing that those
functions are actually static members. This change
makes it more obvious to the compiler, which functions we actually refer
to.
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>
This feature unfortunately relies on a non-standard feature
that is supported by gcc as an extension but triggers
errors when build with clang.
The fix is to disable the feature until we find a better solution
to allow virtual builds with clang.
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>
While older GCC versions accept using an unnamed struct, gcc 7+ does not. Use a
named one instead, to make both gcc versions happy.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Many build systems allow C and C++ compilers to be specified
via environment variables C and CXX.
This commit enables this for kaleidoscope-builder.
Furtheron, on unixoid systems virtual builds are possible through a
command line similar to
CXX=<path to C++ compiler> C=<path to C compiler> ARCH=virtual make
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>
This implements a new device plugin, to drive the Dygma Raise. A few helpers are
also introduces, which are used by the Raise only for now, but are generic
enough so that eventually, other boards may use them too.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
This commit adds a new template hook exploreSketch() that allows plugins to efficiently
obtain compile-time known information about the sketch.
The hook is called before setup and is passed a _Sketch
template parameter that wraps two more types Plugins and StaticKeymap, whose static constexpr methods can be used
to explore the registered plugins and the static keymap.
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>
Hook methods can now be templated.
A template parameter type list, a list of template parameters and
a list of dummy template arguments have been added to the
macro arguments used in _FOR_EACH_EVENT_HANDLER.
Non-template hooks pass empty parenthesis for the three newly
introduced macro arguments.
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>
This enables compile time exploration of the keymap
from all code that is part of the sketch's compilation unit.
Signed-off-by: Florian Fleissner <florian.fleissner@inpartik.de>