Gribouille 0.3.0: A Grammar of Graphics for Typst

Most of the useful conversation landed on fit, not syntax trivia. People who know ggplot liked the direction because layered plotting and declarative mappings are a real upgrade over classic “one function per chart type” libraries like matplotlib. At the same time, several readers questioned how much legacy ggplot naming should be carried over into a fresh API, especially names like `aes` and `labs` that make sense only if you already know R. The bigger energy was around Typst itself. Supporters see it as the first credible attempt in years to replace TeX’s macro-heavy, brittle model with something saner to script and style. Skeptics were not disputing that Typst is nicer. They were pointing out the hard wall of adoption: universities, publishers, and internal document stacks are built on decades of LaTeX packages, templates, and habits, and some key distribution points like arXiv still shape what is practical. A more grounded consensus emerged that Typst is strongest on greenfield work, on teams willing to own their templates, and on documents where programmability and layout quality actually matter. A second fault line was scope. Several comments pushed back on the idea that Typst should replace markdown, arguing that markdown wins because it stays readable as plain text and avoids a compile step. Others argued that markdown collapses once documents need real structure, scripting, typography, data-driven figures, or reusable components. The practical boundary people kept circling was simple: markdown remains fine for README-sized text, while Typst starts to make sense for books, long documentation, papers, posters with structured content, and generated reports. There were also a few implementation questions. Because Gribouille is written in pure Typst rather than using a WebAssembly backend, some readers wondered about performance on large documents or many figures. Another asked about interactive SVG output, and the answer was that Typst can emit SVG, but only standard SVG, not browser-dependent embedded HTML tricks. Overall the mood was enthusiastic about the direction, but with clear recognition that plotting alone will not decide whether Typst breaks out. Ecosystem depth, export targets, and institutional compatibility will.