Looking into this a bit more deeply, this isn't a great drop-in replacement because ESP_LOGx assumes each call is a full "log message" and the macros in Debug.h assume they can piecemeal the log entry parts using printf semantics. Replacing them in this way generates 3 lines for every log (header, message, newline).

I also didn't previously realize that log levels were actually supported in Debug.h they just aren't reflected out to Platform.h. The best direct replacement here is esp_log_write(), but we still lose the level data.

@matth-x do you have any suggestions on the best way to modify the logging infrastructure here to cleanly insert ESP_LOGx with support for levels in the platform layer?

We would also love to have this, as we then tunnel ESP-IDF logs into our web interface.

Any way we can support to make this happen?

Thanks @razvanphp, we do the same. Our current workaround is to use MO_CUSTOM_CONSOLE and mocpp_set_console_out() to provide a custom logging function that uses esp_log_write(), but there are still gaps. Mainly being able to pass forward the log level.

Either way, I may patch this to use esp_log_write instead since it's a better replacement for printf, but we would still need a maintainer to review and approve.

@devunwired Thanks for proposing a better solution to this!

In the current debug macros, the log levels handling is a pain and it will be improved in the next MO major version. The current possible workarounds are:

• Forward everything as "info" message to the ESP logger library
• Do a string match and if the log message starts with [MO] info, then forward the following as info message

Neither of the workarounds are great. I would tend to leave it to the implementer to choose the workaround and make the log messages work on their platform. And otherwise, encourage the MOv2 upgrade. Do you agree, or do you think that the esp_log_write() function is a clear benefit for the general MO user?

In MOv2, it will be possible to set the debugCb2() which also passes the log level to the callback:

I see that the built-in log function makes the same mistake as in the current version and doesn't make use of the ESP logger library. Noted on my end!

I just bumped this change to use esp_log_write(), but if this is handled differently in 2.0 then it may just be informational.

I would tend to leave it to the implementer to choose the workaround and make the log messages work on their platform. And otherwise, encourage the MOv2 upgrade. Do you agree, or do you think that the esp_log_write() function is a clear benefit for the general MO user?

IMO, the callback mechanism makes sense for implementers using a platform that doesn't have built-in support or if you have custom processing to do on the log messages.

Since the library has a platform layer with built-in support for ESP-IDF, then esp_log_write() is the proper implementation of printf() for that platform; and it aligns with the way MicroOcpp is using log functions (multiple calls per line to build a message).

Note: There is also a esp_log_writev(), in case that integrates better with the logging framework.

Okay, agreed. Thanks for your improvement to MicroOCPP, Dave!