Skip to content

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog and this project adheres to Semantic Versioning.

0.5.0 - 2026-04-08

Added

  • File download support in the embedded JupyterLab webview (notebook export, etc.) via QWebEngineProfile download hook with a save-file dialog

Fixed

  • Server startup crash when fps_nbconvert is installed — exclude nbconvert module to avoid pulling in tornado/SSL

0.4.1 - 2026-04-05

Fixed

  • Crash on macOS when creating a QWebEngineProfile — use the default profile instead of constructing a new one, which triggers a fatal assertion in Chromium's ProfileAdapter on Qt 6.10

0.4.0 - 2026-03-25

Added

  • Kernel interrupt support (interrupt_request handler)
  • input() / stdin support for interactive prompts
  • Non-blocking kernel dispatch (execute requests no longer block the Qt event loop)
  • Auto-create python3 kernel spec so JupyterLab finds a kernel on all OSes, with fallback to user Jupyter data dir when sys.prefix isn't writable

Fixed

  • Server startup timeout now raises TimeoutError instead of silently proceeding
  • Server thread restores working directory after shutdown (os.chdir is process-wide)
  • Bumped PySide6>=6.8 (6.5 segfaults on Python 3.12+)
  • Bumped ipython>=8.14 / >=9.2 (Python 3.14 compatibility)

  • Ci

    install libegl1 on Linux, use QT_QPA_PLATFORM=offscreen, drop Python 3.15-dev

  • Resolved pre-existing ruff lint and ty type checker errors

0.3.1 - 2026-03-19

Fixed

  • NotImplementedError: Implement enable_gui in a subclass when switching matplotlib backends
  • All repository URLs now point to the SciQLop org instead of personal fork

0.3.0 - 2026-03-19

Added

  • Rich display support: display() calls produce display_data iopub messages with base64-encoded binary MIME types
  • Matplotlib inline backend auto-activation with backend-aware flush hook (switching to Qt backend stops inline rendering from closing figures)
  • Matplotlib Qt backend (module://jupyqt.matplotlib.backend): native Qt figure windows with zoom/pan toolbar, created on the main thread via MainThreadInvoker
  • ipywidgets/comm protocol support: comm_open, comm_msg, comm_close, comm_info_request handlers enabling interactive widgets
  • Example apps and notebooks for matplotlib and ipywidgets

Fixed

  • QtProxy now recursively wraps non-callable QObject attributes
  • Matplotlib Qt backend lifecycle: proper Gcf cleanup on window close, guarded destroy() against deleted C++ objects

0.2.0 - 2026-03-19

Added

  • cwd parameter on EmbeddedJupyter to set the JupyterLab file browser root directory
  • Lazy server start — start() only starts the kernel thread; the jupyverse server launches on first widget() or open_in_browser() call

Fixed

  • All ruff lint errors resolved across source and test files
  • Release workflow only triggers on tag pushes (no more spurious failures)
  • Documentation deployment added to release workflow

0.1.0 - 2026-03-18

Added

  • Embed JupyterLab in PySide6 applications — no ipykernel, no ZMQ, no qasync
  • Background-thread kernel with IPython InteractiveShell and asyncio event loop
  • Jupyter wire protocol handler over anyio memory streams (execute, complete, inspect, is_complete, history, shutdown)
  • jupyverse kernel plugin with FPS module integration
  • Server launcher managing jupyverse lifecycle in a background thread
  • QtProxy for safe cross-thread access to Qt objects from notebook cells
  • JupyterLabWidget (QWebEngineView) with loading placeholder and ready signal
  • EmbeddedJupyter public API: push(), wrap_qt(), widget(), start(), shutdown()
  • Top-level await support in notebook cells via run_cell_async()
  • Minimal smoke test example (examples/minimal_app.py)
  • Demo app with exposed UI widgets and pre-loaded notebook (examples/demo_app.py)

Fixed

  • All shell access (complete, inspect, is_complete) dispatched to kernel thread
  • IPython completions wrapped in provisionalcompleter() context manager