The default `onSetup` will call `.begin`, to support initializing plugins using
the V1 plugins while using `KALEIDOSCOPE_INIT_PLUGINS`. However, plugins that
implement a compatibility layer so that they can be used with both the new API,
and with `Kaleidoscope.use()` will have a `.begin` method too. Which the default
`onSetup` will call, and we'll register the compatibility layer too, in addition
to the new-style event handlers. This results in many things running twice,
which leads to all kinds of problems.
For this reason, override `onSetup`, so that it does not call `begin`. When used
with `Kaleidoscope.use()`, the plugin will still work, so compatibility is
maintained. But the bug is now gone.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
The default `onSetup` will call `.begin`, to support initializing plugins using
the V1 plugins while using `KALEIDOSCOPE_INIT_PLUGINS`. However, plugins that
implement a compatibility layer so that they can be used with both the new API,
and with `Kaleidoscope.use()` will have a `.begin` method too. Which the default
`onSetup` will call, and we'll register the compatibility layer too, in addition
to the new-style event handlers. This results in many things running twice,
which leads to all kinds of problems.
For this reason, override `onSetup`, so that it does not call `begin`. When used
with `Kaleidoscope.use()`, the plugin will still work, so compatibility is
maintained. But the bug is now gone.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Unfortunately, the way we reported the use of an old-style API also triggered
when not using it. Change that to only trigger when we DO use the old API, by
marking the `combo_t` constructor private.
Unfortunately, this does not allow us to use a custom error message.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Instead of iterating through all the bits, use `__builtin_popcountl()`, provided
by gcc, which should be considerably more efficient.
Fixes#27.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
`Consumer_SNapshot` really should have been `Consumer_Snapshot`, so lets fix it.
However, the typo'd name is left in place as an alias, in order to not break any
existing user code, however unlikely the use of this key is.
Sadly, we can't easily add a deprecation warning, because key_defs.h, which
defines `Key`, depends on `key_defs_consumerctl.h`, so we'd end up with a
circular dependency if we tried to add a deprecation.
Fixes#339.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
As per UPGRADE.md, remove `Kaleidoscope.setup(KEYMAP_SIZE)`,
`event_handler_hook_use`, `loop_hook_use`, `USE_PLUGINS`, `MOMENTARY_OFFSET`,
`key_was_pressed`, `key_is_pressed`, `key_toggled_on`, and `key_toggled_off`.
These were deprecated between July and October 2017, and have been marked for
deletion for over a month.
Also updated UPGRADE.md, moving the section about these to a new, "deprecated
and removed" section.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
When using the old API, fail with a helpful error message that points to
`UPGRADING.md`, which explains in simple detail the migration process.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
The old `RxCy` macros were recently changed to be key indexes instead of
per-hand bit indexes, and `KeyboardHardware.isKeyswitchPressed()` and
`KeyboardHardware.pressedKeyswitchCount()` were introduced as a way to peek into
the keyswitch state. This little change migrates TestMode to use them.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Instead of using a list of left/right-hand states and an overrideable global
callback, use a map of action and key-list pairs. This makes the plugin much
more portable, does not require any hardware-specific knowledge within the
plugin, and does not require us to treat the hands separately.
This in turn, results in a friendlier user interface, at the cost of limiting
the maximum length of a combination to five keys. A small price to pay.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Instead of `getKeyswitchStateAtPosition`, which is long, unintuitive and feels
wrong too, introduce `isKeyswitchPressed`, shorter, better, more
reasonable (because it returns a bool - we support only two states anyway!).
Additionally, add `pressedKeyswitchCount()`, which returns the number of key
switches pressed.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
This is sufficiently low-level that it is OK to use `KeyboardHardware` for it.
It's not a HID thing, either.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Move `KEY_INDEX` here, and turn it into a `constexpr` function, `keyIndex`, and
convert the `RxCy` macros to constexpr values instead.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
To allow the hardware plugin to use a more efficient way of representing the
index (if need be), and to be able to turn it into a constexpr, move it to the
hardware plugin.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
To be used by the hardware implementations, `KEY_INDEX` tells us the index of a
key, from which we can figure out the row and column as needed. The index starts
at one, so that plugins that work with a list of key indexes can use zero as a
sentinel. This is important, because when we initialize arrays with fewer
elements than the declared array size, the remaining elements will be zero. We
can use this to avoid having to explicitly add a sentinel in user-facing code.
Additionally, we add `getKeyswitchStateAtPosition` to the HID facade. See its
documentation in `Kaleidoscope-Hardware`.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Implement `getKeyswitchStateAtPosition`, a hardware-agnostic way to peek into
the keyswitch state. Also transition the `RxCy` macros to `KEY_INDEX`, to make
it easier for hardware with more keys than 64 to implement them, and to make
their values unique across a keyboard, not just across a keyboard half.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Instead of only having an ifdef in the body of deprecated methods, and leaving
it up to the compiler to optimize out the empty & unused, explicitly wrap the
declaration of them within an ifdef too. This will make it easier to remove
everything V1 at a later point, and we're not at the mercy of the compiler,
either.
Fixes#327.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
We do not want user code having to deal with KeyboardHardware, so wrap
.detachFromHost and .attachToHost ourselves.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
These functions can be used to detach from the host, then re-attach, possibly
with different properties, without having to reboot the device.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Having it at the brightest uses too much power, and may result in power use
surges when switching between LED modes and NumPad, which in turn can force the
operating system to disable the whole device. To avoid this, lower the
brightness to 160, a carefull tuned value, also used by the `solidRed` mode in
the factory firmware.
Fixes#9.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
The multiplication by 99 can overflow an int8_t depending on whether the
expression is coerced to a regular-sized int (based on the 99 literal).
Just to be safe, perform a cast to int16_t in case the coercion does not
happen.
Previously, diagonal movements were not reduced in the two axes,
resulting in movement that was too quick. This commit divides diagonal
movements by sqrt(2) / 2 to correct the movement speed.
The net result is that diagonal movement should feel smoother and speed
more as expected.
This removes WakeupKeyboard, because a similar feature was integrated into
BootKeyboard itself, rendering it useless.
See keyboardio/KeyboardioHID#35 for an explanation of the whole situation.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Instead of repeating the same cRGB reading code in three places, use the
recently introduced Focus.readColor() function. This both makes the code
cleaner, and more than 50 bytes smaller too.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Instead of using a lambda (which is not constexpr in C++11), use a temporary,
anonymous struct instance to wrap the `static_assert`, which is constexpr in
C++11.
Fixeskeyboardio/Model01-Firmware#53. Thanks to @noseglasses for finding the
cause, and explaining the fix!
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Many plugins use timers, and most of them will call `millis()`, which isn't
wrong, but isn't the most efficient either. While `millis()` isn't terribly
expensive, it's not cheap either. For most cases, we do not need an exact timer,
and one updated once per cycle is enough - which is what `.millisAtCycleStart()`
is. Having a timer that is consistent throughout the whole cycle may also be
beneficial.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
The member functions of the union now take all arguments const, meaning
that it is not possible to modify this value somehow. This reduces the
chance of subtle bugs and widens the contexts in which these member
functions can be used.
Furthermore, one signature took a `Key' by value while all functions
take `Key' by reference. For the sake of consistency, this was adapted
to.
Making the member functions of Key `const' explicitly flags that they
will not change the union. This will allow to use Key in const contexts.
Adding the `constexpr' specifier to the function makes it possible to
rely on the results at compile time. This puts some kind of restrictions
on the function, especially when using C++11 and not a newer standard,
but these restrictions were already fulfilled, so this seems to be safe.
By far the most common deprecation will be the event handler and loop hook
deprecation. Make them less scary, and point out that unless one's a developer,
they likely need not care.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Moving the deprecation messages to a separate header, and adding a few helpers
allow us to write much more detailed deprecation messages, without needlessly
making the code look incredibly messy.
This also updates most of the deprecation messages to be much more helpful, and
provide hints at how to fix the warnings produced by them.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Due to the plugin API redesign, plugins that migrate may want to ensure they are
compiled against a recent enough Kaleidoscope. Others may opt to provide
separate implementations for each version. For this to work, we need to bump the
API version.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Fixed a conditional so that the event handlers of old-style plugins will be
called. Without this, they don't, and old-style plugins that install event
handlers, would not work.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
With this redesign, we introduce a new way to create plugins, which is easier to
extend with new hook points, provides a better interface, uses less memory, less
program space, and is a tiny bit faster too.
It all begins with `kaleidoscope::Plugin` being the base class, which provides
the hook methods plugins can implement. Plugins should be declared with
`KALEIDOSCOPE_INIT_PLUGINS` instead of `Kaleidoscope.use()`. Behind this macro
is a bit of magic (see the in-code documentation) that allows us to unroll the
hook method calls, avoid vtables, and so on. It creates an override for
`kaleidoscope::Hooks::*` methods, each of which will call the respective methods
of each initialized plugin.
With the new API come new names: all of the methods plugins can implement
received new, more descriptive names that all follow a similar pattern.
The old (dubbed V1) API still remains in place, although deprecated. One can
turn it off by setting the `KALEIDOSCOPE_ENABLE_V1_PLUGIN_API` define to zero,
while compiling the firmware.
This work is based on #276, written by @noseglasses. @obra and @algernon did
some cleaning up and applied a little naming treatment.
Signed-off-by: noseglasses <shinynoseglasses@gmail.com>
Signed-off-by: Jesse Vincent <jesse@keyboard.io>
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Use a timeout calculation method that is not affected by overflow, and also
requires 16 bits less.
This likely fixes#8.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Ticks depend on the speed of the main loop, and as such, are not a reliable way
to time animations. For this reason, use proper timers instead.
The update delay is set to 40ms, which appears to be a slow, relaxing animation,
and should be roughly in the ballpark the tick-based timing was, before speeding
up the main loop considerably.
Fixes#3.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
This warp mode is similar to the navigation provided by some speech
recognition software. A 9-cell grid may provide more precision and
efficiency than the existing 4-cell warp mode. This adds some extra
key definitions to support the additional sectors and enables a user
to switch the grid size:
MouseKeys.setWarpGridSize(MOUSE_WARP_GRID_3X3);
Signed-off-by: Cy Rossignol <cy@rossignols.me>
With the previous algorithm, once every 65 seconds, there would be a significant jump in
the brightness of the "breathing" LEDs as the 16-bit value recorded from `millis()`
overflowed. Instead of dividing by 12, I changed it to a bit shift (4 bits; equivalent to
division by 16), so when the integer overflow occurs, the next value is what it should be.
This is a somewhat unwieldy fix for all the out of bounds (attempted) array addressing at
both ends. When `pos` goes out of bounds in either direction, the test is the same because
it's an unsigned integer. However, after the change of direction, the trailing LED will
still be out of bounds, so we check that every time we call `setCrgbAt()` for `pos2`.
It's rather ugly, but it does ensure that we don't call `setCrgbAt()` with an
out-of-bounds address.
Before this change, we couldn't use the full functionality of the plugin's
warp feature to drag an item (by holding down a mouse button key). The
plugin would reset the warp state during each scan cycle, so we could
only warp the pointer to a cell in the top-level grid. This fix enables
warping repeatedly into sub-cells while holding a mouse button.
This prevents an insignificant error, but it is more correct to handle the integer
overflow instead of ignoring it. I've also changed syncTimer from a 32-bit to 16-bit
integer, which results in a smaller code size, and changed the computation of the timeout
slightly, so the LED update interval is always the same (we add `syncDelay` to the
previous update's start time, not it's end time), rather than varying based on when
LEDControl's `loopHook()` function is called relative to the last timeout.
There already exist 2 rainbow LED effects, this adds a third, using
the LED-Stalker effect.
When you press a key, the LED on that key will cycle through all the
colors of the rainbow, independent of the colors of other keys.
Since keyboardio/Kaleidoscope-Hardware-Model01#23 we do not call
`handleKeyswitchEvent` for keys that are idle. Document this in the comments.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
When a keyswitch has been off in the previous cycle and is still off now, do not
call `handleKeyswitchEvent` on it. As an extension, when the whole column is
idle, skip the whole thing.
In practice, handling fully idle keys is not useful. There are many plugins
which explicitly look for this case and return early, because it isn't an
interesting event. As such, not calling the event handler in this case makes
sense, as we save not only a few needless checks in plugins, but our performance
improves greatly too.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
In `actOnMatrixScan`, there is no need to use temporary variables, we can just
pass the data directly to `actOnHalfRow`, and doing so makes the code easier to
follow.
In `actOnHalfRow`, we can further optimize things if instead of reading the Nth
bit, we always read the first, and shift the byte at the end.
All of these optimizations were done by @obra, I just wrote the commit message.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
When there are no state changes, and no keys pressed on a row, instead of
iterating through a byte bit-by-bit, just fire idle events without checking the
bits. In all other cases, do the bit-walking like we did before.
The reason this is useful is because bit-walking is costly, and slow. If we can
avoid that, we win quite a lot of performance. Since rows being idle is the most
common case on a keyboard, this is a huge net win. Even in the worst case, where
no rows are idle, this is just one byte comparison and a branch slower than our
previous implementation.
As part of this optimization, `actOnHalfRow` was lifted out into its own
function, to reduce code duplication.
Many thanks to @gedankenexperimenter for the original idea!
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Allow user to specify custom boot greeting key by Key_* or by specific row and column. Add ability to define custom duration of boot greet breathing effect, and add ability to change color hue of breathing effect. Finally, rework logic that happens when plugin is loaded to allow all user custom settings to be properly read and applied as expected.
add hue to the header
Updated readme to support new features
astyle + change to allow custom settings
This avoids the OS being stuck with Numlock on, which leaves the LED on
if other keyboard are attached, and breaks FVWM.
See keyboardio/Model01-Firmware#42
These macros were copied and pasted, I'm guessing from the USB HID Usage Tables document,
where this keycode is identified with a parenthetical, unlike all of the others around it:
`AC Download (Save Target As)`. When automatically converted into a preprocessor macro, it
gets a trailing underscore and an argument, which is not what we want.
This change won't actually fix anything that's currently broken, but it might matter
someday.
Reset the bad key bitfields each time a test is started, so that each test starts from a clean slate. Without this, it was confusing to restart the test and get an ever-increasing number of keys to appear bad immediately.
Store any macros key events and play them after the event handler pass has finished, so we
don't have a problem when holding other keys that are handled after the macro key in a
pass. This fixes the problem where held modifiers wouldn't be applied to macros, and also
fast repeating of printing characters.
This change does introduce a limit (default: 8) on the number of concurrent macros that
can be played.
We are moving towards including the Adaptor from the Hardware library, so we
need not pull them in from user sketches (or from core).
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Instead of pulling it in from the user sketch, do so from the hardware plugin.
The hardware and the adaptor are in close relationship anyway, and with
tweakable knobs, we do not need to use a different adaptor library in advanced
cases, either.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Instead of implementing the HID adaptors within Kaleidoscope, provide an API
only (by marking the symbols `extern`). For the sake of backwards compatibility,
pull in `Kaleidoscope-HIDAdaptor-KeyboardioHID`, a new shim library implementing
the status quo.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Instead of using `Keyboard.begin` directly, use
`kaleidoscope::hid::initializeKeyboard`. While there, also initialize
`ConsumerControl` and `SystemControl` the same way.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Use `static_assert` instead of `#error` to report an API mismatch, resulting in
a much more informative error message.
Thanks to @cdisselkoen for the request, and @noseglasses for the `static_assert`
idea!
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
As we guarantee backwards compatibility throughout a major version, it helps if
we have that version available, so plugins / sketches can check if they are
compatible, and issue a helpful error if they are not. As further convenience,
defining `KALEIDOSCOPE_REQUIRED_API_VERSION` before including `Kaleidoscope.h`
will result in a check being done by Kaleidoscope, and an error printed if it
does not match the API shipped.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
As this is a generic plugin, for keyboards that do not have LEDs, don't tie it
to LEDControl, and don't provide a `toggleLEDs` method. Instead, show an example
how to achieve the same thing from the sketch.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
At least on Linux, for a device to be considered capable of waking the host up,
it must be a boot keyboard. As we do not (yet) support a boot keyboard, we fake
one. An USB node that does nothing else than report itself as a boot keyboard,
and does the minimum amount of work to get recognised as such.
Because of this, Linux - and hopefully the other OSes too - will consider the
whole device capable of waking up the host.
This addresses keyboardio/Kaleidoscope#237, if all goes well.
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
Add LEDControl.paused, which we use to pause LED mode updates if true (defaults
to false). This is useful when we want to stop LED modes from updating without
switching to another (like when the host goes to sleep, and we want to turn LEDs
off).
Signed-off-by: Gergely Nagy <algernon@keyboard.io>
This doesn't change behaviour at all; it's just a different way to do the computation,
which I think is much clearer. I also added an explanatory comment.
* It's now all bitwise operations, without arithemetic thrown in.
* It uses the same exact formula for finding bits on both sides of the keyboard.
* It saves 14 bytes in program memory.
The boolean wasOn was unnecessary, and there was no need to call
bitSet() (or bitClear(), in the case of Layer.off()) if the test
passed. Mostly, I just added a few explanatory comments.
(Aslo reversed the sense of the on/off test in Layer.on() and .off())
@algernon likes it better this way, and I agree.
`updateActiveLayers()` makes it impossible to turn off the default
layer, so there's no point searching past it for the highest layer,
and `defaultLayer` can be set to numbers higher than zero.
Now that `layer_count` is (potentially) available, we can start
looking for active layers at the top _defined_ layer instead of the
top _possible_ layer. This ought to be more efficient, especially for
sketches that don't have lots of layers defined.
Also introduced the `MAX_LAYERS` constant (#define).
Instead of discrete press & release tracking, press the mousekey when the
physical key is pressed, and like `Keyboard`, send & clear the report once per
cycle, instead on every action.
This fixes#10.
Signed-off-by: Gergely Nagy <kaleidoscope@gergo.csillger.hu>
Instead of sending a press & release for the consumer key when the physical key
is released, send a press each cycle it is held, a report along with the
keyboard report, and clear the consumer report each cycle too.
This will prevent these keys getting stuck, or sending multiple presses in the
same report.
Fixes#176.
Signed-off-by: Gergely Nagy <kaleidoscope@gergo.csillger.hu>
If we call updateLiveCompositeKeymap() on key release the keymap gets
updated before the release event occurs, and any ShiftToLayer(N) key
with a different definition on layer N won't work properly. Before its
release event is processed, it gets updated to the new value, and
layer N doesn't get turned off. If we only update the live keymap on
key press events, we don't have this problem.
replacing the Key_16bit macro with CONSUMER_KEY macro allowed us to
add the IS_CONSUMER and SYNTHETIC flags within the CONSUMER_KEY macro
and simplify the Consumer key definitions.
If it's an old sketch, LayerCount will default to 0, so in order for
Layer.on() to function, don't bother checking for out-of-bounds if
LayerCount == 0.
Declaring LayerCount as a weak symbol in layers.cpp lets us override
it if the CREATE_KEYMAP macro is used to define the keymap in the
sketch file, but still allows old sketch files to compile without
errors.
Still some changes necessary to allow old sketches to work
properly (Layer.on() will abort before doing anything).
This macro allows the definition of the LayerCount variable and the
keymaps[] array together. It shouldn't break old sketches, but this is
probably not all that's necessary; LayerCount still doesn't get
initialized outside the macro.
Update key_events.cpp (IS_INTERNAL Handling)
I'm merging this for now, even though I know it's not the 'right' solution. But I'd like Mute to work correctly for MP2 keyboards and we're on deadline
This file is meant to be included in sketch files in order to make
data available to Kaleidoscope functions. In particular, the size of
the keymaps[] array (i.e. the number of defined layers), which is
needed in order to prevent reading uninitialized memory past the end
of that array due to Key_KeymapNext_Momentary.
By moving the IS_CONSUMER flag to B00001000 instead of
B00000010 (swap with IS_INTERNAL) we can detect the if the key is a
consumer key and strip out the flags and use the full 10bit to send to
the hid report. This enable us to use all the Consumer_* keys
Reordered if chain in handleSyntheticKeyswitchEvent to fix a bug preventing some Consumer and System Control HID functions from being sent due to bit overload/collision with the IS_INTERNAL flag.
Gergely Nagy
Jesse Vincent
Jack Zhou
Aaron Christianson
Gergely Nagy
Max Görner
noseglasses
Jesse Vincent
Ryan Anderson
Cy Rossignol
Michael Richters
Toon Claes
Marty Gentillon
Ben Gemperline
Maxime de Roucy
Evan Danaher
noseglasses
tazlor
Iliana Weller
Nate Soares
Florian Fleissner
matt venn
SjB
Gergely Nagy
James N. V. Cash
ejinks4
Javaru