Skip to content

Releases: brunobrown/flet-pkg

Release list

v0.2.0

Choose a tag to compare

@brunobrown brunobrown released this 27 May 04:23
429a242

Flet 0.85 Alignment & Generation Quality

Major quality release: generated extensions are now aligned to Flet 0.85.x and compile far more cleanly. The generated Dart remains a reviewable scaffold; advanced cases (required callbacks, complex SDK config types) are documented TODOs.

Highlights:

  • Flet 0.85.x wiring: Extension extends FletExtension with createWidget/createService, widget class {Control}Control in lib/src/, flet>=0.85.1
  • Generated extensions compile much cleaner (−68% flutter analyze errors; simple services = 0 errors)
  • Dart bridge fixes: correct await/sync calls, stream events via .listen(...), buildWidget on widget.control, non-null coalescing (widgets, numbers, strings, enums)
  • Typed reads: ft.ColorgetColor(context), Flutter/package enums → parseEnum(...), typed List/Map args → .cast<T>()
  • Smarter analysis: @visibleForTesting/@protected and example-app classes filtered out; widget-family detection requires shared params (no more over-grouping)
  • Example app wiring fixed: [tool.flet.dev_packages] + [tool.uv.sources] (editable) + [tool.flet] metadata
  • Clear post-creation workflow: build & test with flet build (not flet run) — resolves issue #6
  • New docs: architecture, pipeline, data-model and Python↔Dart bridge diagrams (Mermaid)

v0.1.1

Choose a tag to compare

@brunobrown brunobrown released this 06 Mar 02:58

Initial Release

CLI tool that scaffolds Flet extension packages from Flutter packages, with auto-generated Python controls, Dart bridge code, type mappings, and event handlers.

Highlights:

  • flet-pkg create with interactive and non-interactive modes
  • Auto-detect extension type (Service or UI Control) from Dart source
  • Deterministic generation pipeline: download → parse → analyze → generate → write
  • ~95-100% API coverage on tested packages (shared_preferences, battery_plus, onesignal_flutter, local_auth, image_picker, and more)
  • Sub-module grouping, stream event detection, compound widget detection, platform interface resolution
  • Optional AI refinement (flet-pkg[ai]) using Architect/Editor pattern with Ollama, Anthropic, OpenAI, or Google
  • Optional MCP server (flet-pkg[mcp]) for AI agent integrations (Claude Desktop, Cursor, VS Code)
  • Verbose mode (-v) with coverage score and Rich breakdown table
  • Name conflict detection (PyPI, GitHub, Flet SDK)
  • Python reserved word sanitization in generated code