C++ SDK reference (server-side)

Overview

This topic documents how to get started with the server-side C++ SDK, and links to reference information on all of the supported features.

Prerequisites and dependencies

To use the C++ SDK, you must have the following prerequisites installed on your build machine:

• Windows or a POSIX environment (Linux, OSX, BSD)
• cmake, version 3.19 or above
• boost, version 1.81 or above
• openssl, version 1.1 or above
• libpthread, if you are using a POSIX environment

To build the C++ SDK, you must have the following dependencies. These are automatically fetched by cmake during the build process:

• tl/expected
• djarek/certify

If you are planning to run the C++ SDK test suite, you will also need the following:

• nlohmann/json
• googletest

You do not need to run the test suite in order to use the SDK.

Get started

After you complete the Getting started process, follow these instructions to start using the LaunchDarkly SDK in your C application.

The following sections explain how to install and configure the SDK, and then to verify its connection to LaunchDarkly by fetching flag configuration information.

Incorporate the SDK

You can incorporate the SDK by building from source using cmake, or by using pre-built artifacts. Then, include the LaunchDarkly headers.

Incorporate the SDK using cmake

To incorporate the SDK using cmake:

1. Clone the GitHub repository as a subdirectory of your project.
2. Update your project's CMakeLists.txt to include the SDK repository:
   Using add_subdirectory

   add_subdirectory(cpp-sdks)
   COPY

3. Link your project's target against the launchdarkly::server target:
   Linking your target

   Linking your target, if you are using Redis

   target_link_libraries(your-target PRIVATE launchdarkly::server)
   COPY

Incorporate the SDK using prebuilt artifacts

The C++ (server-side) SDK releases include 64-bit static and dynamic libraries for Linux, Mac, and Windows.

To incorporate the SDK using prebuilt artifacts:

1. Download the correct release for your platform from the GitHub Releases page.
2. Ensure the SDK's headers are installed on the build system. One way to do this is to clone the GitHub repository and install the headers using cmake:
   Install headers

   cmake --build .
   cmake --install .
   COPY

You can now reference the installed headers and link against the prebuilt libraries.

Include the LaunchDarkly headers

To include the LaunchDarkly SDK headers:

#include <launchdarkly/server_side/client.hpp>
COPY

The C wrapper is included in the release binaries.

After cloning the repository, include the LaunchDarkly SDK headers:

#include <launchdarkly/server_side/client.hpp>
COPY

About the SDK namespaces

SDK components common to the C++ (server-side) SDK v3.0 and the C++ (client-side) SDK v3.0 exist within the top-level launchdarkly namespace. Server-side components exist within launchdarkly::server_side.

Individual server-side config builders, for example the LoggingBuilder, are within the launchdarkly::server_side::config::builders namespace. To reduce verbosity of configuration code, you may bring this namespace into scope.

To keep the examples in our documentation concise, we assume symbols in the top-level launchdarkly namespace are visible. You can bring launchdarkly, launchdarkly::server_side, launchdarkly::server_side::config::builders, or all of these into scope, or you can refer to SDK components by their fully-qualified names.

For example:

using namespace launchdarkly; # omitted in examples; assumed to be present
auto config_builder = server_side::ConfigBuilder("sdk-key-123abc");
auto config = config_builder.Build();
COPY

Initialize the client

After you install and import the SDK, create a single, shared instance of Client. To create a client instance, you need your environment's SDK key to authorize your application to connect to a particular environment within LaunchDarkly.

Here's how to configure the SDK key and create the client:

auto config_builder = server_side::ConfigBuilder("sdk-key-123abc");
auto config = config_builder.Build();
if (!config) {
   /* an error occurred, config is not valid */
}
server_side::Client client(*config);
COPY

To learn more about the specific configuration options available for this SDK, read ConfigBuilder.

Next, construct the client and call StartAsync to initiate a request to the LaunchDarkly service and fetch the feature flag settings.

Because the client initializes asynchronously, you may choose to block until the client is ready.

To block on initialization:

client.StartAsync().wait_for(std::chrono::seconds(10));
COPY

To initialize asynchronously:

client.StartAsync();
COPY

Whether you block on initialization or initialize asynchronously, you can also examine the result to determine if initialization succeeded. Here's how:

server_side::Client client(*config);
auto start_result = client.StartAsync();
if (auto const status = start_result.wait_for(maxwait); status == std::future_status::ready) {
    /* The client's attempt to initialize succeeded or failed in the specified amount of time. */
    if (start_result.get()) {
        /* Initialization succeeded. */
    } else {
        /* Initialization failed. */
    }
} else {
    /* The specified timeout was reached, but the client is still initializing. */
}
COPY

If you block on initialization, the initialization call blocks up to the time maxwait. If you request a feature flag before initialization completes, you will receive the fallback value you defined in your variation call.

Evaluate a context

Now you can check which variation a specific context should receive for a given feature flag. Here's how:

auto context = ContextBuilder().Kind("user", "user-key-123abc").Name("Sandy").Build();
bool show_feature = client.BoolVariation(context, "flag-key-123abc", false);
if (show_feature) {
    // application code to show the feature
} else {
    // the code to run if the feature is off
}
COPY

Shut down the client

Lastly, shut down the client when your application terminates.

In the C++ SDK v3.0, the SDK will be automatically closed. If you are using the C binding, or if you are working with earlier versions of the SDK, you must specifically close the client. To learn more, read Shutting down.

Supported features

This SDK supports the following features:

• Anonymous contexts and users
• Configuration, including
  • Application metadata configuration
  • Service endpoint configuration
• Context configuration
• Evaluating flags
• Flag evaluation reasons
• Flushing events
• Getting all flags
• Identifying and changing contexts
• Logging configuration
• Monitoring SDK status
• Offline mode
• Private attributes
• Reading flags from a file
• Relay Proxy configuration
  • Using proxy mode
  • Using daemon mode
• Sending custom events
• Shutting down
• Storing data
• Test data sources