chore(cmake): adopt submodule-first dependency strategy

- Adopt a "submodule-first, FetchContent-fallback" approach for third‑party deps.
- Add and pin submodules under 3rdparty: spdlog, fmt, httplib, skywalking-data-collect-protocol, grpc (pinned tags included).
- Move auto-detection and conflict validation into each dependency module (cmake/*.cmake) and simplify top-level CMakeLists.txt.
- Update CI (main.yml) to checkout submodules and to build gRPC from grpc.
- Add skywalking.cmake and wire proto2cpp to use SKYWALKING_PROTOCOL_PATH.
- Update README.md with submodule workflow, build instructions and bump procedure.

Rationale: improves reproducibility, makes dependency handling explicit and easier to maintain.

Signed-off-by: Matthieu MOREL <[email protected]>

Author: mmorel-35

.github/workflows/main.yml

@@ -80,6 +80,9 @@ jobs:
       - run: echo "/opt/llvm/bin" >> $GITHUB_PATH
 
       - uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
+        with:
+          submodules: 'recursive'
+          fetch-depth: 0
 
       - name: Generate dependencies hash
         id: deps-hash
@@ -90,19 +93,12 @@ jobs:
           echo "hash=$DEPS_HASH" >> $GITHUB_OUTPUT
           echo "Dependencies hash (first 8 chars): $DEPS_HASH"
 
-      - name: Checkout skywalking-data-collect-protocol
-        uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
-        with:
-          repository: apache/skywalking-data-collect-protocol
-          ref: ${{ env.SKYWALKING_PROTOCOL_REF }}
-          path: 3rdparty/skywalking-data-collect-protocol
-
       # Cache gRPC build
       - name: Cache gRPC build
         uses: actions/cache@v3
         with:
           path: |
-            grpc/build
+            3rdparty/grpc/build
             /usr/local/lib/libgrpc*
             /usr/local/lib/libprotobuf*
             /usr/local/lib/cmake/grpc
@@ -115,21 +111,12 @@ jobs:
           restore-keys: |
             grpc-cmake-${{ runner.os }}-${{ steps.deps-hash.outputs.hash }}-
             grpc-cmake-${{ runner.os }}-
-
-      - name: Checkout grpc
-        uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
-        with:
-          repository: grpc/grpc
-          ref: ${{ env.GRPC_VERSION }}
-          path: grpc
-          submodules: true
-
       - name: Install cmake dependencies and run cmake compile
         run: |
           sudo apt-get update
           sudo apt-get install -y cmake build-essential
-          sudo cmake -S ./grpc -B ./grpc/build
-          sudo cmake --build ./grpc/build --parallel 8 --target install
+          sudo cmake -S ./3rdparty/grpc -B ./3rdparty/grpc/build
+          sudo cmake --build ./3rdparty/grpc/build --parallel 8 --target install
           sudo cmake -S . -B ./build
           sudo cmake --build ./build
 

.gitignore

@@ -48,7 +48,6 @@ coverage_report
 
 # CMake build directories and dependencies
 build/
-3rdparty/
 grpc/
 
 ### Automatically added by Hedron's Bazel Compile Commands Extractor: https://github.com/hedronvision/bazel-compile-commands-extractor

.gitmodules

@@ -0,0 +1,19 @@
+[submodule "3rdparty/fmt"]
+	path = 3rdparty/fmt
+	url = https://github.com/fmtlib/fmt.git
+
+[submodule "3rdparty/grpc"]
+	path = 3rdparty/grpc
+	url = https://github.com/grpc/grpc.git
+
+[submodule "3rdparty/httplib"]
+	path = 3rdparty/httplib
+	url = https://github.com/yhirose/cpp-httplib.git
+
+[submodule "3rdparty/skywalking-data-collect-protocol"]
+	path = 3rdparty/skywalking-data-collect-protocol
+	url = https://github.com/apache/skywalking-data-collect-protocol.git
+
+[submodule "3rdparty/spdlog"]
+	path = 3rdparty/spdlog
+	url = https://github.com/gabime/spdlog.git

3rdparty/fmt

@@ -10,13 +10,14 @@ project(cpp2sky
     HOMEPAGE_URL "https://github.com/SkyAPM/cpp2sky"
 )
 
-option(OVERRIDE_CXX_STANDARD_FLAGS "Force building with -std=c++11 even if the CXXFLAGS are configured differently" ON)
-option(SPDLOG_FETCHCONTENT         "Using spdlog FetchContent to build" ON)
-option(FMTLIB_FETCHCONTENT         "Using fmt FetchContent to build" ON)
-option(HTTPLIB_FETCHCONTENT        "Using httplib FetchContent to build" ON)
 option(CPP2SKY_INSTALL             "Generate the install target." OFF)
+option(FMTLIB_FETCHCONTENT         "Using fmt FetchContent to build" ON)
 option(GENERATE_CPP2SKY_PKGCONFIG  "Generate and install pkg-config files for UNIX" OFF)
-
+option(GRPC_FETCHCONTENT           "Using gRPC FetchContent to build" ON)
+option(HTTPLIB_FETCHCONTENT        "Using httplib FetchContent to build" ON)
+option(OVERRIDE_CXX_STANDARD_FLAGS "Force building with -std=c++11 even if the CXXFLAGS are configured differently" ON)
+option(SKYWALKING_FETCHCONTENT     "Using skywalking-data-collect-protocol FetchContent to build" ON)
+option(SPDLOG_FETCHCONTENT         "Using spdlog FetchContent to build" ON)
 
 if(OVERRIDE_CXX_STANDARD_FLAGS)
     set(CMAKE_CXX_STANDARD 17)
@@ -46,6 +47,7 @@ include(fmtlib)
 include(spdlog)
 include(httplib)
 include(grpc)
+include(skywalking)
 include(proto2cpp)
 
 target_link_libraries(${PROJECT_NAME}

3rdparty/grpc

@@ -33,15 +33,92 @@ cc_binary(
 
 #### Cmake
 
-You can compile this project, according to the following steps:
+You can compile this project in one of three supported ways (submodule-first is recommended):
+
+- Recommended — Submodule-first (most reproducible):
+
+```bash
+git clone --recurse-submodules [email protected]:SkyAPM/cpp2sky.git
+git submodule update --init --recursive
 ```
-step 01: git clone [email protected]:SkyAPM/cpp2sky.git
-step 02: git clone -b v9.1.0 https://github.com/apache/skywalking-data-collect-protocol.git ./3rdparty/skywalking-data-collect-protocol
-step 03: git clone -b v1.46.6 https://github.com/grpc/grpc.git --recursive
-step 04: cmake -S ./grpc -B ./grpc/build && cmake --build ./grpc/build --parallel 8 --target install
-step 05: cmake -S . -B ./build && cmake --build ./build
+
+This repository pins several third-party dependencies under `3rdparty/` (example: `spdlog`, `fmt`, `httplib`, `skywalking-data-collect-protocol`). The top-level CMake will auto-detect these submodules and use them via `add_subdirectory()`.
+
+- FetchContent fallback (automatic clone at configure time):
+
+If a submodule is not present, CMake can automatically download the dependency at configure time using FetchContent. This is controlled by CMake options of the form `-D<LIB>_FETCHCONTENT=ON` (these are `ON` by default for some deps). For SkyWalking you can enable FetchContent with:
+
+```bash
+cmake -DSKYWALKING_FETCHCONTENT=ON -S . -B build
+cmake --build build
 ```
 
+- Explicit path (developer workflow):
+
+If you already have a local checkout of `skywalking-data-collect-protocol`, point CMake to it:
+
+```bash
+cmake -DSKYWALKING_PROTOCOL_PATH=/path/to/skywalking-data-collect-protocol -S . -B build
+cmake --build build
+```
+
+Notes about dependencies with special handling
+
+Most third-party dependencies follow the same pattern: prefer the pinned submodule under `3rdparty/` (submodule-first) and fall back to `FetchContent` at configure time when the submodule is not present. See the top-level `CMakeLists.txt` and the modules in `cmake/` for details.
+
+There are two cases worth calling out because their consumption/build steps differ slightly:
+
+- gRPC
+
+  gRPC is not header-only and typically needs configuration and a build step (or its targets must be available via `find_package`). This repository includes a pinned copy of gRPC at `3rdparty/grpc` (tag `v1.74.1`) so builds are reproducible. CI is configured to build gRPC from that submodule.
+
+  To build gRPC locally from the submodule and install it for the rest of the project:
+
+  ```bash
+  # initialize submodules (if you haven't already)
+  git submodule update --init --recursive
+
+  # install build deps
+  sudo apt-get update
+  sudo apt-get install -y cmake build-essential
+
+  # configure & install gRPC from the submodule
+  cmake -S 3rdparty/grpc -B 3rdparty/grpc/build
+  cmake --build 3rdparty/grpc/build --parallel 8 --target install
+
+  # configure & build cpp2sky
+  cmake -S . -B build
+  cmake --build build --parallel $(nproc)
+  ```
+
+  If you prefer not to use the submodule you can still clone and build gRPC separately and make its CMake targets available to the project, but using the pinned submodule is recommended for reproducibility.
+
+- skywalking-data-collect-protocol (protobufs)
+
+  The SkyWalking protocol repository contains the protobuf definitions used to generate code for the project. The `cmake/skywalking.cmake` module sets `SKYWALKING_PROTOCOL_PATH` when the `3rdparty/skywalking-data-collect-protocol` submodule is present so the proto generation step can locate the `.proto` files.
+
+  You can override that behavior by supplying an explicit path:
+
+  ```bash
+  cmake -DSKYWALKING_PROTOCOL_PATH=/path/to/skywalking-data-collect-protocol -S . -B build
+  cmake --build build
+  ```
+
+  Alternatively, enable FetchContent for SkyWalking with `-DSKYWALKING_FETCHCONTENT=ON` to let CMake fetch the proto repo at configure time.
+
+How to bump a submodule (example for skywalking-data-collect-protocol):
+```bash
+cd 3rdparty/skywalking-data-collect-protocol
+git fetch --tags
+git checkout tags/v10.4.0    # or a specific commit
+cd ../..
+git add 3rdparty/skywalking-data-collect-protocol
+git commit -m "Pin skywalking-data-collect-protocol to v10.4.0"
+git push
+```
+
+If you prefer CI to always fetch the latest submodules, ensure the workflow initializes submodules (this repo's CI uses `actions/checkout` with `submodules: 'recursive'`).
+
 You can also use find_package to get target libary in your project. Like this:
 ```
 find_package(cpp2sky CONFIG REQUIRED)