output settings not being retained following reboot in 14.7.2

[wompslab]

What happened?

Issue where output settings revert to default after reboot

Tested on ESP32-D0WDQ5 rev.3 (WLEDMM esp32_4MB_S.bin)

Also reported on ESP32-S3.

To Reproduce Bug

To reproduce:

• Upgrade (fresh install not tested) to 14.7.2 (in this case it was an OTA upgrade from upstream wled v16)
• Change some output settings
• Save
• Reboot

Expected Behavior

This performs as expected until reboot, at which point settings will have reverted to 30x led.

Other config changes (e.g. wifi settings) behave as expected. Config json shows new settings when viewed in UI.

Install Method

From MoonModules Release Page

What version/release of MM WLED?

ESP32-D0WDQ5 rev.3 (WLEDMM esp32_4MB_S.bin)

Which microcontroller/board are you seeing the problem on?

ESP32-D0WDQ5 (tested by me)
ESP32-S3 (mentioned in Discord by another user - not replicated by me)

Relevant log/trace output

https://discord.com/channels/700041398778331156/1031139855427706961/1512983218582720627

Anything else?

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[Funkton]

Same issue here. My Claude says this is a bug introduced in commit b06d3c0 (May 29). The commit was meant to skip the memory check for large RAM boards, but it accidentally wrapped the entire busses.add() call inside the #if, silently discarding the LED config on every boot for regular ESP32 and S3.

[cfg.cpp:205]
The #if guard on line 205 skips adding the bus configs entirely for regular ESP32 on boot:

// This block is compiled OUT for regular ESP32:
#if !defined(ARDUINO_ARCH_ESP32) || defined(CONFIG_IDF_TARGET_ESP32S2) || defined(CONFIG_IDF_TARGET_ESP32C3)
  if (mem <= MAX_LED_MEMORY) if (busses.add(bc) == -1) break;
#endif

For ESP8266, S2, and C3, busses are added from cfg.json correctly. For a regular ESP32, the bus config is read from the file, parsed into a local variable, then silently discarded. When beginStrip() runs, it sees 0 busses and falls back to your compiled-in DATA_PINS/PIXEL_COUNTS defaults - which is why the count resets on every reboot.

The settings page shows what was saved to cfg.json (which is correct), but /json/info shows what's actually running (the compiled defaults).

The fix is to remove the #if guard so regular ESP32 loads busses from the filesystem like every other platform:

#if !defined(ARDUINO_ARCH_ESP32) || defined(CONFIG_IDF_TARGET_ESP32S2) || defined(CONFIG_IDF_TARGET_ESP32C3) // WLEDMM only for boards with small RAM
        if (mem <= MAX_LED_MEMORY)
#endif
        if (busses.add(bc) == -1) break;  // finalization will be done in WLED::beginStrip()

[softhack007]

@Funkton @wompslab thanks for your report 👍

i'll check what happens on esp32. Not sure mr AI is right, but i'll take a look.

[Funkton]

With the fix on my fork it built and corrected the issue. I don't have any of the other boards to test the original fix for the memory check.
Thanks for looking into it