Commit graph

3 commits

Author SHA1 Message Date
Brenno de Winter
2d8be6f0dd Add project docs, EUPL licence, and open-source licence check
Some checks failed
CI / Format · Analyze · Test (push) Has been cancelled
CI / Format · Analyze · Test (pull_request) Has been cancelled
Documentation & licensing:
- Add the EUPL-1.2 licence (LICENSE.md) and set the project licence; refresh
  the README (name origin wink, updated feature list, documentation index).
- Add CONTRIBUTING, SECURITY, CODE_OF_CONDUCT, CHANGELOG, AUTHORS, and
  THIRD_PARTY_NOTICES, plus docs/ (ARCHITECTURE, BUILD, USER_GUIDE, SHORTCUTS,
  LICENSE_COMPLIANCE) and .github/ (CI workflow, issue/PR templates).
- Bring docs/FILE_FORMAT.md in line with current behaviour (code & chart
  slides, per-slide TLP comment, annotation .ink.json sidecar, chart data/ CSVs).

Open-source compliance:
- Add tool/check_licenses.dart and a `make licenses` target (wired into
  check-full and CI) that verifies every resolved dependency uses a recognised
  open-source licence. A scan of all 151 packages and bundled assets found only
  OSI-approved licences.

Charts (Fase 1.1):
- Replace the chart CSV textarea with an in-app editable data grid (editable
  series/labels/values, add/remove row & column, read-only when linked).
- Centralize the linked-CSV directory name (`data/`) in a shared constant.

Also normalize formatting repo-wide with `dart format` and fix one
curly-braces lint, so `make check` and CI are green.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-07 12:19:56 +02:00
Brenno de Winter
227abf351e Add annotation layer (laser, pen, highlighter) over slides
Draw on slides while presenting, kept as a layer fully separate from the
Marp content so the deck stays pure, portable Marp.

Presenter tools (D pen, T highlighter, E eraser, X laser, C clear, Esc
puts the tool away) with a small floating colour/tool bar shown only when
a tool is active. Strokes use coordinates normalized to the 16:9 slide so
they render identically on the laptop and the beamer; in dual-screen mode
the ink and the laser are mirrored live to the audience window.

Persistence (decoupled from the .md):
- In memory the layer is keyed by Slide.id (stable within a session).
- On disk it lives in a sidecar <name>.ink.json next to the deck and as a
  separate entry inside the .ocideck package; the markdown is untouched.
- Because slide ids are regenerated on load, the sidecar anchors strokes
  by order + a content fingerprint, re-attaching them after reordering and
  dropping them when a slide's content changed.
- Deck.annotations carries the layer in memory but is never serialized to
  markdown; deckProvider.setAnnotations keeps it out of undo/redo.

flutter analyze is clean, all tests pass (incl. new stroke/codec tests),
and the macOS debug build compiles. Drawing and live beamer sync still
need verification on real hardware.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-07 11:14:51 +02:00
Brenno de Winter
2aca44365a Add dual-screen presenter mode (slide on beamer, notes on laptop)
When a second display is connected (macOS), presenting now opens a
borderless audience window on the beamer showing the slide, while the
main window shows the presenter view (current/next slide, speaker notes,
clock, controls) on the laptop. The two windows stay in sync over method
channels: navigation, blank screen, audio-complete and beamer clicks are
forwarded between them, and media plays only on the beamer to avoid
double audio. Falls back to the existing single-window presenter when
there is one display or the second window can't be created.

- Vendors a fork of desktop_multi_window in third_party/ that re-adds the
  native macOS window geometry/fullscreen calls (coverScreen, setFrame,
  close) the published 0.3.0 dropped; wired via a path dependency.
- Registers the app's plugins for sub-windows in MainFlutterWindow so
  video/image rendering works on the beamer.
- Routes the multi_window dart entrypoint to a minimal AudienceWindowApp.

Compiles (flutter analyze + macOS debug build) and all tests pass;
runtime two-screen behaviour still needs verification on real hardware.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-06 21:25:34 +02:00