Installation

Getting the source code

This library handles the communication with the DeepAffex Cloud services and companion to the DeepAffex Extraction SDK. It implements the REST and WebSocket protocols necessary for performing a DeepAffex measurement as documented by the DeepAffex API website.

git clone https://github.com/nuralogix/dfx-api-client-cpp.git

Dependencies

There is a long list of dependencies which CMake uses find_package() to locate so these libraries could be injected from another system location, but it is strongly recommended to use conan to provide these dependencies as described in the next section on Build Environment Setup.

Library	Source	Needed for
libcurl	https://github.com/curl/curl	Base
zlib	https://github.com/madler/zlib	Base
fmt	https://github.com/fmtlib/fmt	Base
nlohmann_json	https://github.com/nlohmann/json	Base
openssl	https://github.com/openssl/openssl	Base
yaml-cpp	https://github.com/jbeder/yaml-cpp	Base
libwebsockets	https://libwebsockets.org	WebSocket
Protobuf	https://github.com/protocolbuffers/protobuf	WebSocket & gRPC
gRPC	https://github.com/grpc/grpc	gRPC
googleapis	https://github.com/googleapis/googleapis	gRPC
abseil	https://github.com/abseil/abseil-cpp	gRPC
re2	https://github.com/google/re2	gRPC
meson	https://github.com/mesonbuild/meson	gRPC
c-ares	https://github.com/c-ares/c-ares	gRPC
Doxygen	https://github.com/doxygen/doxygen	Documentation
CLI11	https://github.com/CLIUtils/CLI11	dfxcli tool
naturalsort	https://github.com/scopeInfinity/NaturalSort	dfxcli tool
gtest	https://github.com/google/googletest	Testing
gflags	https://github.com/gflags/gflags	Testing

Build Environment Setup

Note

Building requires a fully C++17-compliant compiler.

This package relies on the dependency management from Conan and the CMake build system. It needs both Conan and CMake to build the library, but can be built by either Conan or CMake. Deciding if Conan or CMake should build the library is related to how you will consume the library.

Conan is a great tool for managing dependencies and if you are writing an application which depends on this library and using Conan already, it is trivial to add as an external dependency which handles the cross-platform and version updating for you.

If you are working on this library, while you could build with Conan it would be a very slow work flow. Conan will cache the build when it’s built so dependent applications don’t need to continuously rebuild, but when working on the library you make a small change, build and test. For this, having a conventional working copy with object files pre-built is more efficient which makes building with CMake a better alternative.

If you are cross-compiling this library, it is easiest to do from Conan which is done by selecting a profile which sets up the appropriate CMake Tool Chain.

Conan/CMake/Just installation is easiest done with a platform package manager, but can be done via source as well:

# OSX (using https://brew.sh/)
brew install conan
brew install cmake
brew install just

#############################################################

# Windows (using https://scoop.sh/)
scoop install conan
scoop install cmake
scoop install just

#############################################################

# Ubuntu
mkdir development
cd development
python3 -m venv venv         # Create Python virtual environment
source venv/bin/activate     # Activate virtual environment
pip install conan

# Install CMake via build (package manager not recent enough)
sudo apt-get install build-essential libssl-dev
wget https://github.com/Kitware/CMake/releases/download/v3.22.3/cmake-3.22.3.tar.gz
tar -zxvf cmake-3.22.3.tar.gz
cd cmake-3.22.3
./bootstrap
make
sudo make install

wget https://github.com/casey/just/releases/download/1.1.0/just-1.1.0-x86_64-unknown-linux-musl.tar.gz
tar xvzf just-1.1.0-x86_64-unknown-linux-musl.tar.gz just
sudo mv just /usr/local/bin

Configuring conan

To retrieve packages which are Nuralogix hosted you will need to setup the Conan remote,

conan remote add nuralogix \
      https://artifactory.na-east.nuralogix.ai/artifactory/api/conan/conan

Quick install instructions

The most basic way to build is using standard Conan install syntax. Expect this to take a while and can be built for both Debug & Release targets.

conan install dfxcloud/1.0.0@ --build missing -rnuralogix -o build_type=Debug
conan install dfxcloud/1.0.0@ --build missing -rnuralogix -o build_type=Release

This places the built artifact into the Conan cache (~/.conan/data) and may take a while if none of the upstream dependencies are available locally or from a Conan server as a pre-built package. Subsequent builds will be much faster, provided your dependency and option graphs remain consistent.

CMake Generators

CMake offers a number of different cmake-generators which can be used to create Ninja build files, Xcode Projects or Visual Studio projects (VS 6 through VS 16 (2019)) from the CMakeLists.txt definitions.

For example, to create a Visual Studio 16 2019 for Win32, x64, ARM or ARM64 you would use one of the following and then open the generated solution files.

cmake -G "Visual Studio 16 2019" -A Win32 ..

cmake -G "Visual Studio 16 2019" -A x64 ..

cmake -G "Visual Studio 16 2019" -A ARM ..

cmake -G "Visual Studio 16 2019" -A ARM64 ..

Troubleshooting build issues

If you force a clean build of everything, ignoring pre-built artifacts it can take about 60 minutes or more depending upon your computer processor speeds.

Sometimes OSX reports errors like: “object file … was built for newer OSX version (10.11) than being linked (10.9)”

Which can be resolved by a rebuild omitting the missing conan build option:

conan install dfxcloud/1.0.0@ --build

Alternatively, you can be more selective and if you know the specific libraries which should be rebuilt.

conan install dfxcloud/1.0.0@ --build openssl --build dfxcloud

Would build both openssl and dfxcloud.