Skip to content

Latest commit

 

History

History
105 lines (74 loc) · 3.3 KB

RELEASE_BACKLOG.md

File metadata and controls

105 lines (74 loc) · 3.3 KB

Release Backlog

This document tracks follow-up work that is intentionally deferred so release preparation can keep moving.

Use it for:

  • release polish that is useful but not blocking a deploy
  • post-release hardening tasks
  • maintainer-facing follow-ups that do not belong in meta/

Do not use it for:

  • frozen architecture decisions
  • implementation discoveries that still affect core direction
  • dependency-ordered implementation planning

If an item becomes active implementation work, move it into issues, a PR, or the appropriate planning document.

Open Items

Package family split after initial single-package release

Status:

  • Deferred
  • Not a release blocker

Why this is here:

  • The current package shape is a good fit for an initial release because it keeps installation and publishing simple.
  • The internal architecture already separates pure core, Cytoscape adapter, and Solid integration concerns clearly enough that a later package-family split is plausible.

Preferred long-term package shape:

  • @munchkeen/core
  • @munchkeen/cytoscape
  • @munchkeen/solid

Why this shape is preferred:

  • It mirrors the existing architectural layers more cleanly than mixed names such as @munchkeen/cytoscape-solid.
  • It lets pure-core consumers avoid renderer and framework-specific dependencies.
  • It gives renderer and framework integrations room to evolve independently.
  • It creates a cleaner package-family story if additional integrations appear later.

Questions to answer before doing this:

  • Whether a convenience package should still exist at munchkeen or whether the package family should become fully scoped.
  • How versioning and release cadence should work across the split packages.
  • Whether the current subpath exports should remain as compatibility shims during migration.
  • How peer dependencies should be divided once Solid and Cytoscape move into separate packages.

Exit condition:

  • Revisit only after the first public release has landed and the single-package distribution shape has proven itself in real usage.

Coverage hardening after first deploy

Status:

  • Deferred
  • Not a release blocker

Why this is here:

  • Current runtime line coverage is 90.8%.
  • The remaining gap is concentrated in a small number of integration and error-path files, not spread evenly across the codebase.

Highest-value follow-ups:

  • Add Graph tests for renderer-session resolution failures, mount failures, apply failures, and the no-op diff path in src/solid/Graph.tsx.
  • Add Cytoscape adapter tests for renderer creation, attach, and stylesheet failure paths in src/renderers/cytoscape/adapter.ts.
  • Add interpreter tests for missing-element and native-layout failure cases in src/renderers/cytoscape/interpreter.ts.
  • Expand branch coverage for deferred edge cases in src/core/project.ts and src/core/transitions.ts.

Notes:

  • Barrel and type-only export files may appear as 0% in reports, but they do not materially drag down the aggregate score because they have no executable lines.
  • Typecheck tests improve safety and public API confidence, but they do not increase runtime coverage numbers.

Exit condition:

  • Revisit after the first deploy stabilizes.
  • Either add the targeted tests above or explicitly trim report noise if a file is intentionally non-executable.