Skip to content

Setting up VSCode w/ clangd

Most of our users use VSCode so we made a guide for them. These guidelines should also work for non-vscode users as well.

Info

The default "C/C++" extension by VSCode is nice that it can figure a ton of things out about your system automatically. But the way we write code for libhal using conan makes the extension difficult and slow to use. This is due to the fact that we need to point the extension to the .conan2/p/ (conan 2 package directory). Because that directory could have a large number of files and multiple version of the same project, it tends to get confused, stop working, or get very slow. We recommend clangd because it is fast, helpful, and easy to use.

Setup Steps

  1. Install VSCode if you don't already have it installed.
  2. Go the the "Extensions" section on the left side bar. Hover over the icons to get their name.
  3. Search for "C/C++" and disable the extension if it is already installed and enabled.
  4. Search for the extension clangd and install the extension.
  5. Install clang
    1. Ubuntu/Debian: sudo apt install clang
    2. MacOS: brew install llvm
    3. Windows: choco install llvm
  6. Add --query-driver=**/arm-none-eabi-g++ (for Linux & Mac) or --query-driver=**/arm-none-eabi-g++.exe (for Windows) to the clangd command arguments in the clangd settings.
    1. Go the the extensions page
    2. Find clangd and press the GEAR icon and open up settings.
    3. Find the settings clangd: Arguments and add the above code there. It is comma delimitated so you can add additional compiler patterns as well.

How clangd works

You are almost done, but we need to discuss what is needed to make clangd work. A workspace will need a compile_commands.json file to be present in your root directory or to use a .clangd file at the root of the repo that configures where to look for the .json file. compile_commands.json tells clangd what commands you are using in order to determine exactly how your files are built and what commands are used to build them, which provides the following benefits:

  1. More accurate warnings and error messages in the IDE
  2. Faster response time because only the necessary includes for the specific version you are targeted will be used in the evaluation.

Enabling clangd

For a libhal library projects

If you are contributing to libhal project/repo, then those libraries and demos will already be using libhal-cmake-util/[^4.0.5] which will automatically enable the generation of a compile_commands.json file. To get this file, run:

conan build .

And it will be generated.

If you are attempting to do with is a demo or an application you will need to specify the platform and compiler like usual.

conan build . -pr lpc4078 -pr arm-gcc-12.3

For your own project

You can either add a self.requires("libhal-cmake-util/[^4.0.5]") to your project or add the following lines to your CMakeLists.txt.

# Generate compile commands for anyone using our libraries.
set(CMAKE_EXPORT_COMPILE_COMMANDS ON)

# Always run this custom target by making it depend on ALL
add_custom_target(copy_compile_commands ALL
    COMMAND ${CMAKE_COMMAND} -E copy_if_different
    ${CMAKE_BINARY_DIR}/compile_commands.json
    ${CMAKE_SOURCE_DIR}/compile_commands.json
    DEPENDS ${CMAKE_BINARY_DIR}/compile_commands.json)

Now run conan build . (where . is the path to your project or library) and it should generate the compile_commands.json file.

Ensure that you include the necessary profiles added to the build.

conan build . -pr stm32f103 -pr arm-gcc-12.3

Refreshing the LSP

Now that you should have your compile_commands.json in the right location, you just need to refresh your LSP.

  1. In VSCode Press: ⌘+shift+P on Mac or Ctrl+Shift+P on everything else.
  2. Select the following command: clangd: restart language server

Now your LSP should be active and your C++ files should be able to find your includes as well as infer the types of your objects.