Machine Learning

简体中文 | English

Machine Learning is a native Linux desktop teaching application for the Machine programming language, written in C with GTK4, GtkSourceView 5, and SQLite3.

This first scaffold already implements:

• Linux-only desktop packaging
• First-launch onboarding wizard
• Instant English / Simplified Chinese interface switching
• User-selected database directory
• Resume-from-last-node persistence via SQLite
• A persistent main window named List1
• Plain .mne editing with line numbers and lesson-aware run/check workflow
• Install / uninstall targets that do not depend on the source tree after installation
• Isolation from the Machine compiler runtime paths under /usr/local/lib/machine and /usr/local/include/machine

Design Notes

• The wizard is fixed-size and non-resizable.
• The main window is the long-lived workspace window called List1.
• If the selected database directory is deleted later, the application returns to the onboarding wizard on the next launch.
• The app installs its own resources into /usr/local/share/machine-learning and does not overwrite or relocate the Machine compiler runtime.
• The code editor keeps plain text editing behavior and does not replace or modify the Machine compiler installation.

Window Size Mapping

GTK windows use pixel sizes, while the original product brief described window sizes as 100x35 and 145x45.

This scaffold maps those layout units to practical desktop defaults:

• onboarding wizard: 1000 x 700
• List1 main window: 1450 x 900

If you want a different mapping later, it is easy to change.

Dependencies

Fedora

sudo dnf install -y gcc make gtk4-devel gtksourceview5-devel sqlite-devel desktop-file-utils

Debian / Ubuntu

sudo apt install -y build-essential libgtk-4-dev libgtksourceview-5-dev libsqlite3-dev desktop-file-utils

Arch Linux

sudo pacman -S --needed base-devel gtk4 gtksourceview5 sqlite desktop-file-utils

Build Commands

Running plain make prints the English build parameter summary:

make

Build the application:

make build

Run it from the source tree:

make run

Install it system-wide:

sudo make install

Uninstall it:

sudo make uninstall

Installed Layout

After installation, the project no longer depends on the source tree.

/usr/local/bin/machine-learning
/usr/local/share/machine-learning/css/app.css
/usr/local/share/machine-learning/language-specs/machine.lang
/usr/local/share/applications/machine-learning.desktop
/usr/local/share/icons/hicolor/scalable/apps/machine-learning.svg

Configuration and Database

User configuration is stored at:

~/.config/machine-learning/config.ini

The actual lesson database is stored inside the folder selected by the user during onboarding:

<selected-folder>/machine_learning.db

What Works Right Now

• first launch language chooser
• immediate UI language switching
• database folder selection
• SQLite bootstrap
• persistence of the current lesson node
• automatic resume to the previous node
• List1 main workspace opening directly on later launches

What Comes Next

This scaffold is ready for the next phase:

• full course graph
• rich lesson rendering
• Machine code execution integration
• progress checkpoints and pass/fail tracking
• more advanced editor tooling
• deeper HTML/CSS-like card layouts and learning modules