This change allows the user to set the log-level with homeSpan.setLogLevel() instead of needing to change in Settings.h. Next up: Update Example that introduces VERBOSE to instead use setLogLevel(). Consider adding an interactive 'L' command to change this during run-time.
Also, added "#pragma once" to every *.h file to prevent duplication of definitions. TO DO: Review all #include to ensure sanity across all files.
Next Up: Add method to break out of WiFi connection loop by using push button in case WiFi SSID/PWD has changed but there is no desire to unpair controllers. To Do: Move blinking status light setting to named macros.
And updated other SRP routines to complete implementation of use of stored verification code instead of live setup code.
Next up: Add method to change setup code in serial interface and implement similar code after network configuration routines.
This generates an SRP verification code from a setupCode and randomly-generated salt. Function creates the salt internally and returns both the resulting verification code and salt that was used. These are stored in NVS permanently.
Next Step: create SRP6A:loadVerifycode()
Move all configuration code into Network.cpp, eliminating any calls or connection to HAP.cpp. Network.cpp is now standalone for HomeSpan.cpp to call directly to handle ALL apects of WiFI and Setup Code initialization.
Also move TempBuffer from HAP.h to Utils.h so it can be used by Network.h
This is no longer needed since PushButtons can be easily emulated from within a Service loop() as per Example 14. Library is now fully up-to-date: TimedResets and SpanEvents have been fully removed.
AND updated Example 13 to reflect new loop() framework in place of SpanEvents. ALSO found a bunch of inconsistencies with WindowCovering HAP documentation. PositonState and HoldPosition are NOT used by HomeKit. However, HomeKit has a full slider for controlling shades which makes a Hold Button no longer needed. See Example 13 for details. Open to do: add commentary to Example 13 and eliminate SpanEvents from library!
Completed new framework for using loop() methods and deleted all code related to new SpanEvent() and event() loops. Re-worked Example 13 to utilize new framework and validated it functions as expected.
Service loops() called for only those Services with over-ridden loop() methods. TO DO: update event logic to follow new framework. Vector should point to all CHARACTERISTICS that are updated to setVal()
This check all timed write PID/TTL entries and clears all those that are expired. This completes the implementation of Timed Write functionality as required for PUT /prepare requests.
converted checkedTimeResets() from two-pass logic that first computes the required size of SpanBuf to a single-pass that utilizes vector<SpanBuf> instead. Much cleaner and easier.
checkEvents() now allows for multiple characteristics to be updated by service->notify(). This change included the use of vector<SpanBuf> to dynamically create the the temporary storage needed. To do: re-work similar routines to use vector<SpanBuf> instead of needing to perform two passes each time to gauge size of SpanBuf array needed.
creating code to allow for multiple characteristics to be updated within event() call. To Do: define anogther sensor that supports multiple characteristics. also need to check that "new SpanEvent" works as expected from within a defined service.
putCharacteristics and checkNotifications now both call the same eventNotify routing. This is being done to prepare for services that will stream notifications back to client (like a temperature sensor, or pushbutton). TO DO: rename checkNotifications to check TimedResets, and rename SpanPut to something more generic.
Allow for any characteristic that is not read-only, which should leave out all string characteristics.
Upon reset, zero out all 8 bytes of UVal memory, instead of simply setting UVal.BOOL to false. This allows use of TimedReset for any characteristic with a numerical value.
Updated SpanRange to register itself automatically to the most recent Characteristic. This is more consistent with the overall framework and no longer requires user to access the range variable directly.
Gregg