- select_effect: clear pixels + paint current layer when entering effect #4
- select_effect: restart tick timer when leaving effect #4 for normal effects
- on: paint current layer when toggling RGB on while on effect #4
Fixes: underglow stuck on old colors, frozen display until layer switch,
cycling effects requiring off/on toggle.
The effect_layer() tick handler was fading all pixels to black every
frame, then calling transient_off() which killed RGB entirely. Layer
colors are static — set by set_layer() on layer change events. No
per-frame animation needed.
In preparation for the re-do of #2036 events that are useful for displays/indicators on the peripheral should be compiled on both sides so they can be raised on both sides
I tried using event instead of the 25ms underglow_tick to update the underglow on
layer change only. That didn't improve the battery life much....
The second change is cutting off the led strip power if the underglow is
not defined for a layer. Power is restored if a layer with rgb is
activated, and cut off as soon as the layer is disabled.
This, on the other hand, improves the battery life a lot, especially if you
don't use rgb on your base layer.
If you are using rgb on your base layer, setting
CONFIG_ZMK_RGB_UNDERGLOW_AUTO_OFF_IDLE is highly recommended.
Adds a Kconfig setting for the default EXT_POWER status at initialization time. Previously it was always initialized to on if no saved value was present.
Adds Glove80's status indicator using RGB underglow support. Requires ZMK
PR#999 and PR#1243.
The underglow status is able to show layer state, battery levels,
caps/num/scroll-lock, BLE and USB state. The underglow positions selected for
each of these indicators is configured using the new devicetree node
zmk,underglow-indicators, which takes an array of integer LED positions for each
feature.
Adds some encoder tests for rotating, and basic layers.
Mock Kscans don't seem to interact nicely, slightly on the hacky side.
However, better to have hacky tests than no tests.
Fixes for regressions from split refactors that broke polling mode
specifically.
Co-authored-by: honorless <86894501+lesshonor@users.noreply.github.com>
Co-authored-by: Cem Aksoylar <caksoylar@users.noreply.github.com>
Remove the previous condition on ZMK_SPLIT_WIRED that only enabled it if
!ZMK_BLE. We'll defer to the `depends on` to ensure the DTS exists to
use wired split to make that decision on enabling the feature.
Allow the nice!view widget to display the status of each profile by:
- not adding a circle around the number if the profile is not bound
- drawing a dashed circle around the number if a profile is bound but not connected
Given a typical setup of a battery at ~4 V and a nice!nano v2 running at
3.3 V, I measured a small increase in power usage (using a Nordic PPK2)
when SOC_DCDC_NRF52X_HV was enabled. This changes the nice_nano_v2 and
mikoto boards so this option is disabled by default.
Fixes#2990
Changed CONFIG_ZMK_STUDIO_TRANSPORT_UART to automatically enable itself
whenever a zmk,studio-rpc-uart chosen node is specified. The previous
behavior of enabling if CONFIG_ZMK_USB was enabled broke builds when
CONFIG_ZMK_STUDIO was enabled but the chosen node wasn't specified.
feat(split): Runtime selection of split transport
Allow building multiple split transports, and select an active
one based on the transport availability. Wired split availability
depends on additional `detect-gpios` which must be a GPIO pin
that goes active when a wired connection is present.
feat(split): Suspend/resume wired UART devices.
To better support runtime split support, suspend/resume the UART
as necessary to save power when not using the UART.
docs(split): Document adjusting nRF52 UART interrupt priorities
For wired split on nRF52, you may need to adjust the priority for UART
interrupts lower, to ensure the interrupts used for timing sensitive
BT operations can run when needed, so document this in our pinctrl docs.
refactor(split): Restore use of aync UART on nRF52.
With fixes for Zephyr UART driver, re-enable using async API on nRF52.
fix(split): Minor wired split fixes.
Various minor fixes for wired split to avoid spurious TX in half duplex,
etc.
fix: Unconditionally define HID payloads to avoid error.
Don't conditionally define HID indicator payload, to avoid compilation
errors.
docs(split): Expand on details of split transports.
Expand the split keyboard documentation with a more fleshed out section
on the available split trasnports, and what is and isn't supported by
each, including the runtime selection functionality.
---------
Co-authored-by: Nicolas Munnich <98408764+nmunnich@users.noreply.github.com>
* Reference combos by index, not 32-bit pointers, and store bitfields
instead of arrays in several places, to bring down our flash/RAM usage.
* Use bit field to track candidate combos, to avoid needing an explicit
`ZMK_COMBO_MAX_COMBOS_PER_KEY` setting.
* Determine the max keys per combo automatically from the devicetree,
so we remove the ZMK_COMBO_MAX_KEYS_PER_COMBO Kconfig symbol.
