Changelog

All notable changes are recorded here. The project is pre-1.0 and follows semantic versioning for development releases.

Unreleased

Renamed the distribution to gsp-vispy2 0.2.0 and the independent producer import to gsp_vispy2; removed the ambiguous vispy2 compatibility package.
Moved the public learning path from the legacy Canvas/RendererBase system to semantic GSP 0.2 records, explicit capabilities, structured results, and explicit execution boundaries.
Removed legacy mesh-shading aliases and tightened transport initialization, sequencing, lifecycle, snapshot identity, command status, and diagnostic contracts.

Consolidated the GSP 0.2 target specification into ten detailed normative chapters with 90 stable requirement identifiers and explicit dispositions for 101 source documents.
Added machine-readable producer and renderer profiles that separate strict, adapted, partial, unsupported, and blocked scopes with concrete evidence.
Added requirement-to-test traceability, a generated public feature matrix, and consistency tools enforced by CI.

Added an executable first tutorial sourced directly from examples/docs/first_scene.py.
Added curated producer, lifecycle/transport, scene/resource/visual, query, and diagnostic API pages.
Added 0.2 migration guidance, legacy URL redirects, screenshot provenance, repository links, and explicit source-only/pre-1.0 maturity boundaries.

Matplotlib remains the required portable reference backend and publishes an exact 0.2 profile.
Datoviz v0.4 remains optional and capability-gated; no symbol, screenshot, or unrelated test is treated as feature promotion.
Added an experimental explicit open_session("datoviz") preview for capability inspection, bounded blocking display, one-frame polling, structured diagnostics, and deterministic cleanup.
Texture2D mesh rendering remains unsupported by Matplotlib and blocked from Datoviz promotion; producer support does not imply renderer support.
No production current-protocol remote transport, general retained display-update API, close callbacks, or event-loop embedding contract is claimed.

Full pytest, targeted strict mypy, Ruff, package build, specification/profile/public-doc checks, executable tutorial, compatibility redirects, and strict MkDocs build.

Initial object-oriented plotting and visualization prototype.
Matplotlib, legacy Datoviz, network, Pydantic, and high-level plotting experiments.
Early examples, tests, and project documentation.