To make the "surpass Tartube" roadmap concrete, document Tartube's
v2.5.231 codebase as a specification we can refer to:
- Module map: 146k lines of Python across 19 modules, with per-module
responsibilities (mainapp / mainwin / config / downloads / media /
options / wizwin / process / refresh / tidy / info / updates /
ffmpeg_tartube / formats / ttutils / dialogue / files / xdg).
- Data model: container hierarchy (Folder → Channel | Playlist → Video)
with N-level Folder nesting, system folders that act as smart filters
(Bookmarks, Favourites, Waiting, New, Live, Missing, Recent,
Temporary), and the full per-video flag inventory.
- OptionsManager: the headline feature — 164 distinct yt-dlp flags
attachable to any container or video with cascading resolution. The
spec lists the option groups (Behaviour, Network, Playlist, Filters,
Download, Filesystem, Auth, Anti-bot, Format, Subtitles,
Post-processing, Cookies, Output, Tartube-specific) and points to a
concrete SQLite + Rust struct design for our Phase 1.1.
- Seven threaded operation managers (Download, Update, Refresh, Tidy,
Info, Process, Livestream/Stream) and three downloader strategies
(VideoDownloader, ClipDownloader, StreamDownloader).
- Custom downloads (named recipes) and profiles (saved container
groups) — feeding our Phase 1 + 3.
- UI surface inventory: five primary tabs (Videos, Progress, Classic
Mode, Drag and Drop, Errors), ~80 specialised dialogs, three
wizards (Setup, Import YT subs, Tutorial), and the ~30-tab
SystemPrefWin.
- SponsorBlock + chapter persistence (slice_list / stamp_list) — a
Phase 1 add-on after per-channel options ship.
- Notable corners: master/slave dbid dedup, RSS-driven discovery,
external_dir per-channel storage, livestream message parsing, full
gettext i18n.
- Things Tartube does better (15 items, mapped to roadmap entries),
things we already do better (10 items, to keep momentum honest).
- A 12-checkbox "Tartube parity" milestone definition — when all are
ticked, the comparison table at the top of ROADMAP.md goes from
8-ahead-8-behind to 8-ahead-0-behind, and "surpass" begins.
ROADMAP.md gains a pointer to the spec just under the north-star line.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Tartube is the mature open-source yt-dlp GUI in this space and the
obvious benchmark. As of today we're 8-ahead / 8-behind / 1-tied across
the comparison matrix at the top of the new ROADMAP.md.
Four phases:
1. Tartube feature parity — the 8 items they have that we don't, ordered
by user-visible impact. Per-channel custom download options is the
single biggest gap; everything else (folder hierarchy, filter UI,
comments capture, notes, system tray, format conversion pipeline,
per-distro packaging) follows.
2. Polish where Tartube is mature — integration tests, docs site,
structured error recovery / "rescue recipes", backup-restore, stability
hardening.
3. Surpass — cross-compile mac+win, Android client, WebSocket-driven
progress, smart auto-tagging, federation, a real comment viewer,
perceptual-hash dedup, plugin/scripting hooks.
4. Stretch / blue-sky — TV-mode layout, AI summarisation, multi-user,
Plex/Jellyfin source-plugin integration.
The architectural wins we already have (single Rust binary, real web
UI, bundled curl_cffi venv, Plex export with NFO sidecars, Argon2 auth +
CSP + rate-limit, 10 themes, ETag+gzip on the library response) are
called out so we keep building on them rather than re-litigating them.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
