January 2024 Summaries

Civo, a cloud-native service provider, has successfully adopted a monorepo strategy by migrating its projects into a single repository. By using Dagger, a CI/CD tool, Civo was able to speed up its build times from 30 minutes to 5 minutes and improve the reliability of its pipelines. The team also gained better visibility into each other's work and reduced operational friction. Additionally, Dagger allowed the team to write pipelines in Go, empowering developers to manage their own pipelines. With this new setup, Civo expects to expand its use of Dagger in the future, including creating multi-language pipelines using Python and Rust SDKs, and leveraging modules to improve its machine learning tools.

The Python, Typescript, and Go SDKs now offer a globally available client instance, dubbed "dag", which can be referred from anywhere, even before establishing a connection. This simplifies code by reducing verbosity and reusability. The global client was introduced in the Python SDK as an experimental feature in v0.8.5 and later adopted as a convention for all SDKs. It provides basic usage through `dagger.connection()`, which establishes a connection in the global client instance, replacing the need to pass a client instance around when making pipelines. The global client also simplifies utility factories by eliminating the need for factory functions that passed a single client instance argument. Additionally, it enables lazy API fields and automatic connection for certain use cases. The Typescript SDK has supported the global client since v0.9.4, while the Go SDK added support in v0.9.6. Adopting the global client will make it easier to adapt to upcoming features like Dagger modules, which manage connections internally. By switching to the global client, your code becomes less verbose and simpler to reuse, making it an attractive option for managing pipelines in non-trivial cases.

Open Source AI is accelerating rapidly, with one of the key advancements being the integration of Project Zenith into Dagger, which enables the use of self-contained, reusable modules for seamless interaction. Luke Marsden, a Daggernaut, has been working on this and has showcased the potential of Dagger for AI use cases in recent demos, including the Helix Module within the Dagger framework. These advancements are made possible by the community's engagement and collaboration, as seen in the Discord Demo Forum where community members can ask questions and learn more about using Dagger for their AI use cases.