December 2018 Summaries

LaunchDarkly, a feature flag service provider, has introduced multivariate feature flags that allow engineers to define custom settings. These feature flags can provide Boolean values or an arbitrary set of string values, number values, or a combination of the three. This new feature opens up possibilities for long-term feature flags and helps drive actual site operations. The author demonstrates this concept by creating a multivariate feature flag that changes the minimum log-level being used by a given application.

In this episode of "Scaling Software", Edith and Paul discuss the evolution of open source communities from benevolent contributors to corporate entities vying for control. They explore how shared goals, ownership, and accountability are crucial in maintaining a successful open source project. Additionally, they touch upon the importance of payment for contributions and the challenges associated with dividing compensation fairly among contributors.

LaunchDarkly has introduced a new feature that allows developers to use file-based flag values for local testing without connecting to the service. This capability is supported in Go, Java, .NET, Ruby, and Python SDKs, with full support for all server-side SDKs coming soon. The files can be standard JSON or YAML format, containing basic information about a flag or more detailed data such as rules, targets, or segments. This feature is intended for local testing, integration testing, and other non-production environments only.

AWS has introduced improvements to their continuous delivery offerings for AWS Fargate and Amazon ECS, allowing users to configure blue/green deployments using AWS CodeDeploy. This enhances Progressive Delivery capabilities. While both blue/green deployment and feature flagging are best practices that minimize risk during infrastructure or application changes, they serve different purposes. Blue/green deployment is an operational practice for changing infrastructure safely, while feature flagging is a development practice for altering any part of the application with minimal risk. Combining these two strategies allows for continuous delivery of value to customers with reduced risk.