Slyde

Slyde is a native wxPython slideshow for editable Python percent notebooks. A slide is introduced by a # %% Title [slide] marker. Python and markdown cells after it belong to that slide until another slide marker appears.

Set up

Install uv and sync the project:

uv sync

On Linux, wxPython may need system GTK and WebKit development/runtime packages. Markdown uses wx.html2.WebView when available and falls back to wx.html.HtmlWindow.

Launch

uv run slyde
uv run slyde examples/interactive.py
uv run slyde examples/animation.py

examples/animation.py builds a square wave out of sine harmonics and uses each of the animate settings described below.

Without a path, use File → Open. A notebook with PEP 723 inline metadata is started through uv run --with for each declared dependency; its packages are made available in the GUI process, so code and embedded Matplotlib widgets share the notebook's environment. uv must be installed and available on PATH. When opening such a notebook from the file picker, Slyde starts a new app process under uv and closes the current window. This first run may resolve and download packages.

The example declares its packages at the top of the notebook:

# /// script
# requires-python = ">=3.11"
# dependencies = ["matplotlib", "numpy"]
# ///

Inline requirements are resolved by uv on demand. No adjacent lockfile is created automatically; add version bounds to dependency strings when a notebook needs tighter version constraints.

Notebook format

# /// script
# requires-python = ">=3.11"
# dependencies = ["matplotlib", "numpy"]
# ///

# %% Introduction [slide]
# %% [markdown]
# Use markdown cells to explain the idea.

# %% [slide]
# %% [markdown]
# Slides may contain more than one cell.

# %% Build a plot
import matplotlib.pyplot as plt

Supported cell types are Python (the default) and [markdown]. [slide] starts a new slide; text before the type marker is its optional title. Marker lines and the preamble are retained when saving. Only Python cells are editable in the UI.

A marker may carry key="value" options after its type. [slide] takes animate:

# %% Results [slide] animate="bullet"

• none shows the whole slide at once, and is the default.
• paragraph reveals each top-level markdown block and each code cell one at a time.
• bullet does the same, and also reveals list items one at a time.

Controls

• Shift+Enter runs the focused Python cell.
• The Kernel menu runs several cells at once or starts over:
  • Run All Cells (Cmd+Shift+A) runs every Python cell in the notebook.
  • Run Until Current Slide (Cmd+Shift+U) runs the Python cells on the slides before the current one.
  • Restart Kernel & Clear All Cells (Cmd+Shift+R) discards the shared namespace and every cell's output, after asking for confirmation.
• Left/Up and Right/Down change slides when focus is outside a code editor, and step through the content of an animated slide before moving on; Cmd+Left/Up and Cmd+Right/Down work even while editing a cell (the platform command key is Ctrl on Windows/Linux).
• Cmd+plus and Cmd+minus scale slide content between 50% and 200% (the platform command key is Ctrl on Windows/Linux).
• F11 toggles full screen; Escape exits full screen.
• Ctrl+S saves edits to the source .py file.

Python cells run in a shared namespace in the GUI process, one at a time. Printed output and exceptions appear below the cell, and long lines scroll horizontally instead of wrapping. Matplotlib figures created by a cell are embedded as interactive wx canvases with the standard toolbar for home, back/forward, pan, zoom, and save. A code cell is as tall as its source, and the plot beside it takes the size Matplotlib needs for it. A plot shrinks by up to 30% when that makes the slide fit the window; otherwise it stays full size and the slide scrolls. Output appears as it is produced, and the window stays usable while a cell runs. A queued run from the Kernel menu shows its progress in a strip at the bottom of the window, and stops at the first cell that raises, moving to that slide.