Scientific documentation precedents¶
Reviewed on 2026-08-25. This note records specific structural decisions; it is not a list of visual brands to imitate.
Learn Astropy¶
Learn Astropy organizes tutorial material around a concrete scientific task, states learning goals, exposes the notebook source, and offers a route for browser execution. We adopt the task-first structure and explicit learning goals. We do not make a notebook runtime the primary reading surface: every RadialPaths tutorial receives a static rendered page first.
Photutils¶
Photutils 3.0 separates getting started material, conceptual user guidance, and factual API reference on its front page. We use the same information hierarchy. The RadialPaths landing page will lead with the two-operation workflow, then link separately to tutorials, coordinate definitions, and API reference.
SunPy¶
The SunPy example gallery groups examples by scientific operation and gives each example a direct, descriptive title. We adopt concise task titles and a visual tutorial index, but limit the initial gallery to four maintained notebooks so it remains easy to scan.
scikit-image¶
The scikit-image gallery pairs rendered results with runnable source and organizes examples by image operation. We adopt the rendered-result preview and source/download pairing. Our examples remain astronomy-facing and distinguish geometry construction from tracer measurement.
Gammapy¶
Gammapy 2.1 provides a short getting-started path,
conceptual user guide, tutorial notebooks, and a distinct API reference. Its
tutorial collection
offers both rendered notebooks and downloadable notebook/source bundles. We
adopt the visible separation between learning material and reference material.
The gallery provides static HTML, notebook downloads, and Colab links pinned
to the immutable v0.1.0 source.
Decisions for this project¶
The landing page explains one workflow: build geometry once, then measure registered tracers on that geometry.
/tutorial/is static HTML and remains useful with JavaScript disabled./explorer/is a bounded browser-native experiment. It demonstrates the coordinate construction without loading Python or a notebook kernel.Four notebooks form the maintained tutorial sequence. Each declares its author, summary, learning goals, keywords, and resources in the first cell.
Tutorial cards show a real rendered result, not a decorative stock image.
API reference stays factual. Explanatory comparisons belong in concepts or tutorials, not in generated signatures.
Navigation labels use astronomer-facing terms: Learn, Explore, Reference, and Cite. “Provenance” is not used as a top-level public label.