lwip | A Lightweight TCP/IP stack | Networking library

I solved this problem. When I changed the following definition that was included in mqtt_connect task, the problem was solved.

You have a naming conflict. You're trying to create a custom component named mqtt which requires the system component named mqtt and you expect the build system to figure it out. That's just asking for trouble :)

Name your component myproject-mqtt or something. Watch out for similar problems with your components wifi and ble.

dsb ish works as a memory barrier for inter-thread memory order; it just orders the current CPU's access to coherent cache. You wouldn't expect dsb ish to flush any cache because that's not required for visibility within the same inner-shareable cache-coherency domain. Like it says in the manual you quoted, it finishes memory operations.

Cacheable memory operations on write-back cache only update cache; waiting for them to finish doesn't imply flushing the cache.

Your ARM system I think has multiple coherency domains for microcontroller vs. DSP? Does your __DSB intrinsic compile to a dsb sy instruction? Assuming that doesn't flush cache, what they mean is presumably that it orders memory / cache operations including explicit flushes, which are still necessary.

Using wordsexp.h solved the issue by parsing the string correctly for getopt().

Essentially:

I have not found the solution to this problem with the mentioned frameworks. Instead I reverted to another dependency from the ESP-IDF: esp-modem. I managed to get a working solution by following the pppos-client and modem-console examples within my current software.

The ESP-IDF build system is built on top of CMake. This means you can use all the standard features of CMake in your files. However, the the ESP-IDF system predefines many functions, and makes many assumptions about the layout of your project, supposedly to make things "easier". Instead of reading CMake documentation, start by reading and understanding the ESP-IDF build system documentation:

https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-guides/build-system.html

It looks to me like there is a particular layout expected for subcomponents, including the format of the CMakeLists.txt file. Specifically, move Metriful under a new directory called components, or add Metriful to EXTRA_COMPONENT_DIRS near the top of your root CMakeLists.txt

If Metriful is not written as an esp-idf component, this may not work. However, the document also describes how to link to "pure CMake" components, which will look something like this (at the end of your root CMakeLists.txt).

extern "C" void logLineToUart(const char * log, ...);

You need to tell it to make the name C-linker compatible the first time it is seen.

The extern "C" is only when compiling as C++. It is not legal syntax in C. You can use extern by itself (it is implied if you leave it off).

That's not very friendly... C++ accommodates sharing header files with C by accepting syntax that would not apply to C++, just to make it easy to use the same header file contents. In particular, using (void) for empty parameter lists (which is "an abomination") and allowing a comma in variadic function parameter lists ( (int x ...) is how C++ originally defined it; when the same feature got incorporated into ANSI C they used (int x, ...) with a superfluous comma that is not needed by the grammar.) C++ compilers accept both of these to make it easier to consume C headers...

But, you have to add extern "C" around everything. This introduces conditional compilation anyway, even though C++ accepts the C syntax. Note that for a single declaration, extern int foo (int); the extern is allowed and implied if you leave it out. If the C compiler allowed the linkage specification even though only "C" is available, it would make life easier. Note that in most cases, the C and C++ implementation are the same compiler suite, and often one language supports some features of the other as extensions. It would be cool if gcc etc. supported extern "C++" in C mode, since that code base certainly does know how the name encodes the parameters.

Application should not assume that esp_http_client_read reads same number of bytes specified in the length argument. Instead, application should check the return value of this API which indicates number of bytes read by corresponding call. If the return value is zero, it indicates that complete data is read.

To work correctly, application should call esp_http_client_read in a while loop with a check for return value of esp_http_client_read.

You can also use esp_http_client_read_response, which is a helper API for esp_http_client_read to internally handle the while loop and read complete data in one go.

Please refer http_native_request example which uses esp_http_client_read_response() API.

What you have pasted are the request headers that your browser sends to the server. The problem is with the response headers that the embedded server sends back. In particular the Content-Type header isn't correct. It should look something like: