* NeighborInfo: Only keep neighbors in RAM
It fills up quickly when nodes are running >=2.3
* Defer first transmission as it's usually empty
---------
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
This changeset gives us the ability to specify a separate SPI device for the LoRa, Display, and Touchscreen. The changes in Portduino also add support for specifying a new SPI speed for each transaction. All together, this means that we can let the Linux OS manage the CS lines, and also get much faster SPI speeds, leading to better framerates.
* Add multiple SPI devices to put Radio, Display, and Touchscreen on each their own
---------
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
The main use case for this will be to create a custom Heltec WiFi LoRa 32 V3 SG_923 variant, which will be pre-flashed and sent for regulatory approval for retail sale.
Signed-off-by: Andrew Yong <me@ndoo.sg>
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* basic identification of TCA9555
* recognise LSM6DS3 on alt address
* keep variant.h changes out of this PR
* 2nd attempt to keep variant.h changes out of this PR
---------
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* fix sleep part 1
* always show wakeup reason in debug log
* fix screen turn on issue
* avoid unnecessary reboot when entering light sleep
* set DIO1 based on radio type
---------
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* started work on pulling in the unphone library and dependencies, to do e.g. power switch management and etc.; currently failing at Adafruit_ImageReader
* now compiles with unphoneLibrary included
* successfully pulled in unphone library to manage power switch and init vibe motor and etc.
doesnt print to serial tho...
* simplified the build a bit; when doing meshtastic do not depend on the MCCI lora libs etc., then also no need to config them via build flags
* version that doesnt trigger brownout
* cleaned up initVariant a little
* note re. GPS
* back to mesh upstream version
* this time we're back to mesh upstream version
* getting LSM6DS3TRC driver installed
* shake to wake works, set threshold quite low may need increasing
* whats the crack with these end of file changes?
* paramatize the wake threshold
* try to get the PR to just include real changes
* got the right config item and also not giving compiler messages
* moved the lib_deps for the LSM6DS3TRC driver from our variant platformio.ini to the main one in root so all boards have it
* stuupid error #define-ing
---------
Co-authored-by: Hamish Cunningham <hamish@gate.ac.uk>
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* Make ButtonThread instance extern
Previously was a static local instance in setup(). Now declared in ButtonThread.cpp, accessible via extern declaration in ButtonThread.
* Extract attachInterrupt() calls to public method; create matching method for detachInterrupt()
* Change suspension of button interrupts for light-sleep
* Fix declaration for ARCH_PORTDUINO
* Remove LOG_DEBUG used during testing
* Don't assume device has a button..
* Guard entire constructor code
* Don't use BUTTON_PIN with ARCH_PORTDUINO
---------
Co-authored-by: Manuel <71137295+mverch67@users.noreply.github.com>
* - map fn+m to mute and unmute the external notification module
- map fn+t to be an alternative for the TAB key
* add whitelist to inputbroker
* (maybe) sweet-talking t-deck into tabbing...
* now for real - back in Kansas
* More fancy mappings
---------
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* Update E-Ink display after sending adhoc ping or disable/enable GPS
* Resume display updates when touch button pressed
* Use touch hold as modifier; change double-click behavior for user button
* Fix preprocessor exclusions
* Purge backlight behavior
* Distinguish between 3x and 4x multi-presses
* Touch button considers "Wake screen on tap or motion" user-setting
* Don't assume device has BUTTON_PIN
* Rename misleading method
* Don't save Neighbors to flash when receiving
* Move `shouldSave` to `saveProtoForModule()`
---------
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* remove redundant logic
* Handle special screens for old EInkDisplayClass
* Handle special screens for EInkDynamicDisplay class
* Join an async refresh in progress to avoid skipping screensaver
* attempt trunk fix
* fix Wireless Paper double-clear screen at boot
* log when flooded with "responsive" frames
* show the "resuming" screen when waking from deep-sleep
* rename drawDeepSleepScreen
avoid future confusion with "Screen Paused" screen
* show a screensaver frame when screen off
The frame shown during deep sleep is now also passed through showScreensaverFrames()
* Add macros for E-Ink color values.
OLEDDISPLAY_COLOR is inverted. Result of light-mode on E-Ink vs dark-mode on OLED?
* adapt drawDeepSleepScreen to new screensaver convention
* Mark Wireless Paper V1.1 as having problems with ghosting
Any other issues can be marked in a similar way, then handled in code where relevant
* Change screensaver from fullscreen logo to overlay
* identify "quirks" rather than "problems"
* move async refresh polling from display() to a NotifiedWorkerThread
* Prevent skipping of deep-sleep screen
(Hopefully)
* Redesign screensaver overlay
Now displays short name
* Optimize refresh for different displays
* Support older EInkDisplay class
* Don't assume text alignment
* fix spelling of a quirk macro
(No impact to code, but avoids future issues)
* Handle impossibly unlikely millis() overflow error
Should have just let it go, but here we are..
---------
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* Use `alterReceivedProtobuf()` for NeighborInfo and Traceroute
`alterReceived()` should never return NULL
Traceroute should be promiscuous
* Remove extensive logging from NeighborInfo module
