Best of Graphite.dev — 2024

1. 1
   Article

   Graphite.dev·2y

   The new developer toolchain

   The post discusses the relevance of Graphite in the era of AI code generation tools like GitHub Copilot and ChatGPT. It argues that AI will make the 'outer loop' of code review, testing, and deployment more critical. With AI-generated code increasing, the need for robust 'outer loop' tools to manage review and deployment workflows is emphasized. The post also explores the limitations and future improvements of AI coding models, stressing the necessity of human oversight in code quality and security.

   143

   4
2. 2
   Article

   Graphite.dev·1y

   How we redesigned Graphite's landing page in-house

   A team redesigned Graphite's landing page in-house to create a compelling and memorable user experience. By experimenting with neon lighting and incorporating user feedback, they achieved a visually distinctive design that aligns with Graphite's brand. The process involved creative engineering to balance visual impact and performance across devices, with a focus on iterative improvement and close collaboration between design and engineering teams.

   36

   1
3. 3
   Article

   Graphite.dev·1y

   AI code generation will remain fragmented

   The code generation market is highly fragmented with no single tool dominating due to the accessibility of foundational LLMs, open standards for code interfaces, and the dispersed nature of context across multiple platforms. This lack of sustainable competitive advantages means developers will continue to use a variety of tools, tailoring their choices to specific tasks and workflows. The dynamic landscape aligns with Hamilton Helmer’s 7 Powers framework, suggesting that the code generation space will remain diverse and competitive without any clear leader.

   24
4. 4
   Article

   Graphite.dev·2y

   How Google migrated billions of lines of code from Perforce to Piper

   Google migrated from Perforce to Piper, their custom-built version control system, over four years to handle the immense scale of their codebase. Perforce's single server was struggling with the load, motivating the shift. This migration was complex due to Perforce's deep integration into Google's engineering environment and included production dependencies. The switch to Piper reduced operational risks and paved the way for new systems. The effort reflects not just Google's resources, but its engineering resilience and innovation.

   20

   2