| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Keyboard: revamp frosty-flake leds
This commit transitions bpiphany/frosty_flake to led_update_{kb,user}
and rewrites the AVR bit twiddling logic to use the standard QMK GPIO
API.
* Keyboard: rewrite frosty_flake's matrix reader to be a lite custom matrix
This commit replaces frosty_flake's custom matrix and debounce logic
with a "lite" custom matrix. In addition to being somewhat clearer, this
allows a consumer of the flake board to choose their own debouncing
algorithm. The one closest to the implementation originally in use is
sym_g, but this opens us up to supporting eager_pk and eager_pr.
The original matrix code was 18 columns for 8 rows, but using a single
row read and unpacking the bits into individual columns. To simplify,
I've changed the key layout to be 8C 18R instead of 18C 8R: this lets us
use a single read directly into the matrix _and_ drop down to a uint8_t
instead of a uint32_t for matrix_row_t.
Since we're no longer implementing our own debouncing and row unpacking,
we save ~400 bytes on the final firmware image.
Fully tested against a CM Storm QFR hosting the flake -- this commit
message was written using the new matrix code.
Firmware Sizes (assuming stock configuration as of 42d6270f2)
Matrix+Debounce Size (bytes)
--------------- ------------
original 17740
new + sym_g 17284
new + eager_pr 18106
new + eager_pk 18204
I expect that there are some scanning speed benefits as well.
* Keyboard: update frosty_flake's UNUSED_PINS
* Keyboard: Remove meaningless weak redefinitions from frosty
These are not necessary (and all of them already live somewhere in
Quantum).
|
