Currently at 1.92-beta, in development!
See UPGRADING.md for more detailed instructions about upgrading from earlier versions (even earlier betas). This is just a list of noteworthy changes.
Better protection against unintended modifiers from Qukeys¶
Qukeys has two new configuration options for preventing unintended modifiers in the output, particularly when typing fast:
Qukeys.setMinimumHoldTime(ms)sets the minimum duration of a qukey press required for it to be eligible to take on its alternate (modifier) value.
Qukeys.setMinimumPriorInterval(ms)sets the minimum interval between the previous printable (letters, numbers, and punctuation) key press and the press of the qukey required to make the qukey eligible to take on its alternate (modifier) value.
KALEIDOSCOPE_API_VERSION has been bumped to 2 due to the plugin API
changes mentioned below. It does not mean that version two of the API is final,
though. The bump is there so plugins can check it, and make compile-time
decisions based on it. Such as whether to compile for the version one, or for
the version two API.
The API version will remain the same, even if we introduce breaking changes - until a stable release is made from the v2 branch. From that point onwards, the API version will change with further breaking changes.
New device API¶
A new hardware device API was introduced in November 2019, replacing the old system. It was designed to be more composable, more future proof than the old system. All hardware plugins under Keyboardio control have been updated to use the new APIs.
See UPGRADING.md for more information.
New plugin API¶
A new plugin API was introduced in May 2018, which replaces the old system. The new system is hopefully easier to use and develop for:
It does not require one to register / use hooks anymore. Implementing the interface provided by
kaleidoscope::Pluginis all that is required.
The new system has more hook points, and the method names are much more clear now.
Plugins under Keyboardio control have all been updated to use the new API, and they no longer support the older one.
See UPGRADING.md for more information.
Transition to a monorepo¶
We heard a lot of complaints that while the plugin architecture of Kaleidoscope is great, having so many plugins scattered around in dozens of repositories is a huge barrier of entry for potential contributors, and a significant pain point for end-users to update. For these reasons and more, we merged almost all plugins into the Kaleidoscope repository.
While at first it may seem that this is a move towards a monolithic architecture, rest assured, it is not. The plugin APIs are still a core part of Kaleidoscope, it isn’t going anywhere. We merely rearranged the sources, is all. Nothing else changes.
Some headers and names did change, however, see UPGRADING.md for more information.
Bidirectional communication for plugins¶
The bi-directional communication protocol formerly implemented by
Kaleidoscope-Focus has been partially pulled into core, using the new plugin system mentioned above. The new system makes it a lot easier for both end-users and developers to use the feature.
See UPGRADING.md for more information.
Numerous plugins use timers, most of them directly calling
millis(). This has the disadvantage that calls within a main loop cycle will be inconsistent, which makes timing synchronization across plugins hard. The newly introduced
Kaleidoscope.millisAtCycleStart() function helps dealing with this issue.
See UPGRADING.md for more information.
USB detach / attach¶
It is now possible to detach, and re-attach the USB link from/to the host, without resetting the device. The intent of this feature (as implemented by the
Kaleidoscope.attachToHost() methods) is to allow configuration changes without rebooting.
See the Kaleidoscope-USB-Quirks plugin for a use-case.
Finer stickability controls for OneShot¶
A way to slow down Unicode input¶
In certain cases we need to delay the unicode input sequence, otherwise the host is unable to process the input properly. For this reason, the Unicode gained an
.input_delay() method that lets us do just that. It still defaults to no delay.
Better support for modifiers in the Cycle plugin¶
The Cycle plugin has much better support for cycling through keys with modifiers applied to them, such as
LSHIFT(Key_A). Please see the documentation and the updated example for more information.
More control over when to send reports during Macro playback¶
There are situations where one would like to disable sending a report after each and every step of a macro, and rather have direct control over when reports are sent. The new
SEND_REPORT steps help with that. Please see the Macros documentation for more information.
LED-ActiveModColor can be asked to not highlight normal modifiers¶
The plugin was intended to work with OneShot primarily, and that’s where it is most useful. To make it less surprising, and more suitable to include it in default-like firmware, we made it possible to ask it not to highlight normal modifiers. Please see the LED-ActiveModColor documentation for more information.
Events now trigger on layer changes¶
Changing layers now triggers the
onLayerChange event - but only if there was real change (thus, calling
Layer.on(SOME_LAYER) multiple times in a row will only trigger one event). This event was introduced to help plugins that depend on layer state schedule their work better.
Hyper and Meh keys¶
To make it easier to create custom shortcuts, that do not interfere with system ones, an old trick is to use many modifiers. To make this easier,
Ctrl+Shift+Alt is commonly abbreviated as
Ctrl+Shift+Alt+Gui is often called
Hyper. To support this, we offer the
Key_Hyper aliases, along with
HYPER(k) to go with them.
keymap internals are now a one dimensional array¶
Historically, Kaleidoscope used the dimensional array
keymaps to map between logical key position and hardware key position.
keymaps has been replaced with
keymaps_linear, which moves the keymap to a simple array. This makes it easier to support new features in Kaleidoscope and simplifies some code
PER_KEY_DATA_STACKED macros are available (when defined by a hardware implementation). These macros make it easier to build features like
KEYMAPS that track some data about each key on a keyboard.
New hardware support¶
Kaleidoscope has been ported to the following devices:
Atreus: All known variants of the original Atreus are supported. From the Legacy Teensy variant, through the pre-2016 PCB with an A* MCU, the post-2016 PCB, and FalbaTech’s handwired one too. Apart from the legacy Teensy variant, the other support both the A* or a Teensy as an MCU.
ErgoDox: Originally developed to support the ErgoDox EZ, but all other compatible hardware is supported, such as the original ErgoDox and anything else wired like it, like some Dactyls.
Planck: AVR-based Plancks, and anything else wired similarly should be supported, as long as they use a Teensy.
Splitography: Initial support for the Splitography Steno keyboard.
For more information, please see the hardware plugins’ documentation.
To make it easier to port Kaleidoscope, we introduced the
ATMegaKeyboard base class. For any board that’s based on the ATMega MCU and a simple matrix, this might be a good foundation to develop the hardware plugin upon.
The DynamicMacros plugin provides a way to use and update macros via the Focus API, through Chrysalis.
The IdleLEDs plugin is a simple, yet, useful one: it will turn the keyboard LEDs off after a period of inactivity, and back on upon the next key event.
The [LEDActiveLayerColor][plugins/Kaleidoscope-LEDActiveLayerColor.md] plugin makes it possible to set the color of all LEDs to the same color, depending on which layer is active topmost.
The Turbo plugin provides a way to send keystrokes in very quick succession while holding down a key.
The WinKeyToggle plugin assists with toggling the Windows key on and off - a little something for those of us who game under Windows and are tired of accidentally popping up the start menu.
Implementation of type Key internally changed from C++ union to class¶
Key was originally implemented as a C++ union. For technical reasons
it had to be converted to a C++ class. This implies that the double usage
of the original union, holding either raw data (member
raw) or key code/key flags
flags) is no more possible.
Direct use of member
emit a diagnostic compiler message but will cause the firmware linking
process to fail. For a deprecation
flags keep on being supported but will cause
deprecation warnings during compile.
Please see the relevant upgrade notes for information about how to upgrade legacy code.
LEDControl.paused has been deprecated¶
.paused property of
LEDControl has been deprecated in favour of the new
LEDControl.enable() methods. These two will turn
off or refresh the LEDs, respectively, along with disabling or re-enabling
future updates and syncs.
NumPad plugin no longer toggles
NumPad plugin used to toggle
NumLock when switching to the NumPad layer. This caused issues on OSX where
NumLock is interpreted as
Clear. For this reason, the plugin no longer does this. As a consequence, everyone’s encouraged to update their keymap so that the numpad layer uses normal number keys instead of the keypad numbers. See Model01-Firmware#79 for an example about how to do this.
RxCy macros and peeking into the keyswitch state¶
RxCy macros changed from being indexes into a per-hand bitmap to being an index across the whole keyboard. This mostly affected the MagicCombo plugin.
Please see the relevant upgrade notes for more information.
Redial plugin had a breaking API change¶
Color palette storage has changed¶
The LED-Palette-Theme had to be changed to store the palette colors in reverse. This change had to be made in order to not default to a bright white palette, that would draw so much power that most operating systems would disconnect the keyboard due to excessive power usage. With inverting the colors, we now default to a black palette instead. This sadly breaks existing palettes, and you will have to re-set the colors.
We also changed when we reserve space for the palette in EEPROM: we used to do it as soon as possible, but that made it impossible to go from a firmware that does not use the plugin to one that does, and still have a compatible EEPROM layout. We now reserve space as late as possible. This breaks existing EEPROM layouts however.
EEPROM-Keymap changed Focus commands¶
The EEPROMKeymap plugin was changed to treat built-in (default) and EEPROM-stored (custom) layers separately, because that’s less surprising, and easier to work with from Chrysalis. The old
keymap.roLayers commands are gone, the new
keymap.custom commands should be used instead.
EEPROMSettings’ version() setter has been deprecated¶
We’re repurposing the
version setting: instead of it being something end-users
can set, we’ll be using it internally to track changes made to
EEPROMSettings itself, with the goal of
allowing external tools to aid in migrations. The setting wasn’t widely used -
if at all -, which is why we chose to repurpose it instead of adding a new
Key masking has been deprecated¶
Key masking was a band-aid introduced to avoid accidentally sending unintended keys when key mapping changes between a key being pressed and released. Since the introduction of keymap caching, this is no longer necessary, as long as we can keep the mapping consistent. Users of key masking are encouraged to find ways to use the caching mechanism instead.
We fixed way too many issues to list here, so we’re going to narrow it down to the most important, most visible ones.
Support for BIOSes, EFI, login prompts, etc¶
Keyboards report keys pressed to the host via either of two protocols: the boot protocol, or the report protocol. The boot protocol is the simpler, and it is what older BIOSes, EFI, and certain OS login prompts (or hard disk password prompts and the like) require. Until recently, the firmware wasn’t able to provide this protocol, only the more advanced report one, which is required for N-key roll-over.
We now support the boot protocol, and on operating systems that fully conform to the USB specification, this works automatically. For all others, one can implement a way to force one mode or the other. See the factory firmware for an example how to achieve this.