fatal error during build

[webmst]

Hi everybody,
I've just cloned the ESP-AT repository on my local machine, executed 'build.py install' (both without any error) but cannot execute the next 'build.py build' due to a fatal error. Here is the standard error output of cmake, clearly showing the error point:

Code: Select all

esp-at branch: master
esp-at docs version: latest
no extra component dirs
Traceback (most recent call last):
  File "C:\esp-at\esp-idf\tools\kconfig_new\confgen.py", line 607, in <module>
    main()
  File "C:\esp-at\esp-idf\tools\kconfig_new\confgen.py", line 328, in main
    output_function(deprecated_options, config, temp_file)
  File "C:\esp-at\esp-idf\tools\kconfig_new\confgen.py", line 370, in write_header
    deprecated_options.append_header(config, filename)
  File "C:\esp-at\esp-idf\tools\kconfig_new\confgen.py", line 176, in append_header
    if new_opt in config.syms and _opt_defined(config.syms[new_opt]):
                                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\esp-at\esp-idf\tools\kconfig_new\confgen.py", line 163, in _opt_defined
    if opt.orig_type in (kconfiglib.BOOL, kconfiglib.TRISTATE) and opt.str_value != 'n':
                         ^^^^^^^^^^^^^^^
AttributeError: module 'kconfiglib' has no attribute 'BOOL'
CMake Error at esp-idf/tools/cmake/kconfig.cmake:176 (message):
  Failed to run confgen.py
  (C:/Espressif/python_env/idf5.3_py3.11_env/Scripts/python.exe;C:/esp-at/esp-idf/tools/kconfig_new/confgen.py;--list-separator=semicolon;--kconfig;C:/esp-at/esp-idf/Kconfig;--sdkconfig-rename;C:/esp-at/esp-idf/sdkconfig.rename;--config;C:/esp-at/sdkconfig;--defaults;C:/esp-at/module_config/module_esp32s2_default/sdkconfig_silence.defaults;--env-file;C:/esp-at/build/config.env).
  Error 1
Call Stack (most recent call first):
  esp-idf/tools/cmake/build.cmake:590 (__kconfig_generate_config)
  esp-idf/tools/cmake/project.cmake:530 (idf_build_process)
  CMakeLists.txt:121 (project)

Does anybody have a clue?
Thanks in advance for the help.

[webmst]

I solved it myself.

As it can be seen on the standard error output, the problem occurred due to the usage of a wrong version for ESP-IDF (v5.3) with ESP-AT instead of the correctly referenced v5.0; that happend because - in the attempt of saving some disk space - I had just deleted the older framework.

Though, as ESP-AT works with its own framework version, I would like to know if Espressif Systems has plans for an upgrade. As we cannot say that the ESP-AT project is kept current despite of its stategic importance, I would also like them to publish a document showing how to add support for not-supported modules (not just changing parameters), at least to allow users to do that themselves.

[ESP_Sun]

Here are some reference materials that might be helpful to you. However, if you are looking to support a completely new module series not yet covered in the AT documentation (such as the ESP32-H series), it could be very challenging. The AT documentation also advises against attempting this, as it involves many difficulties.
1. https://docs.espressif.com/projects/esp ... r-a-module
2. https://docs.espressif.com/projects/esp ... index.html

[webmst]

Thanks for having read my (a bit off-topic) final note about the availability of an ESP-IDF framework upgrade (beyond v.5.0.x) for ESP-AT and the availability of a guide for compiling ESP-AT on every EspressIf SOC, given the importance of this application and the lack of a complete coverage.

I had already read the two suggested documents and I'm currently in the process of adding some custom AT commands on ESP32-S2, as I cannot use an ESP32-S3 (that would be dual-core, provide bluetooth and a lot more of embedded FLASH & PSRAM) on my PCB at the moment.

I can realise it would be *challenging* to add ESP32-S3 support from the customer's side (and let me say - from a developer's viewpoint - that it shouldn't be, either from the supplier's side), but which are the options if EspressIf doesn't declare a roadmap for ESP-AT upgrades?

In my humble opinion, as new SOCs and new ESP-IDF frameworks will be introduced, either EspressIf will have to support an upgraded and more encompassing ESP-AT application or it will have to discontinue its support. As a professional user, I would like to know which one of the previous is more likely to happen, before investing more time in this solution.

[ESP_Sun]

Among the chip platforms supported by AT, both ESP32C3 and ESP32C6 support BLE. Compared to ESP32S3, these two chips offer better cost-effectiveness. I’m curious to know why your product doesn’t use these chip platforms for running AT. Additionally, AT relies on the at_core lib, so customers cannot independently add support for new chips, can only extend support for new modules on chips that are already supported by AT.

[webmst]

Thanks for your interest, from a hardware viewpoint we preferred the ESP32-S2/3 architecture for 2 main reasons: greater GPIO availability and native USB support, moreover the rich integrated FLASH/PSRAM is an extra. Then, from a software viewpoint we chose ESP32-S2 among the two because of the current ESP-AT firmware availability, still compromising on a single I2S module instead of two (this required a software implementation via ISR for the I2S port in CAMERA MODE, even it is fair to say that ESP-IDF v.5.0.x doesn't natively support it). Though - with some hope for the future - our PCB accounts for installing either ESP32-S2 or ESP32-S3 onboard.