Replace HTTPie subprocess calls with native libsoup3 async API. This
eliminates external dependencies and uses GNOME-standard async patterns.
Changes:
- Rewrite http_client.py to use libsoup3 Session and Message APIs
- Replace threading with GLib async callbacks in window.py
- Remove HTTPie dependency from Flatpak manifest
- Update documentation to reflect libsoup3 usage
Benefits:
- No external dependencies (HTTPie removed)
- Native GLib async integration (no threading complexity)
- Better performance with connection pooling and HTTP/2 support
- Cleaner codebase following GNOME development patterns
Use --ignore-installed to ensure all dependencies (including Pygments)
are installed to the Flatpak prefix, not relying on SDK packages that
won't be available at runtime.
- Update history_manager to use GLib.get_user_config_dir()
- Update project_manager to use GLib.get_user_data_dir()
- Remove old filesystem permission from Flatpak manifest
- Fix Flatpak manifest: runtime version 49→47, correct git source URL
Data now stored in:
- Flatpak: ~/.var/app/cz.vesp.roster/{config,data}/cz.vesp.roster/
- Native: ~/.{config,local/share}/cz.vesp.roster/
Roster is a modern HTTP client application for GNOME, similar to Postman,
built with GTK 4 and libadwaita.
Features:
- Send HTTP requests (GET, POST, PUT, DELETE)
- Configure custom headers with add/remove functionality
- Request body editor for POST/PUT/DELETE requests
- View response headers and bodies in separate tabs
- Track request history with JSON persistence
- Load previous requests from history with confirmation dialog
- Beautiful GNOME-native UI with libadwaita components
- HTTPie backend for reliable HTTP communication
Technical implementation:
- Python 3 with GTK 4 and libadwaita 1
- Meson build system with Flatpak support
- Custom widgets (HeaderRow, HistoryItem) with GObject signals
- Background threading for non-blocking HTTP requests
- AdwTabView for modern tabbed interface
- History persistence to ~/.config/roster/history.json
- Comprehensive error handling and user feedback
