48e154d53c
A joining peer learnt the roster by announcing itself and having every member reply with their state. That's one broadcast per member — O(N^2) messages for a room of N — and the in-memory channel layer caps each connection's queue at 100 and drops the overflow silently. Measured with 100 synthetic peers: every peer ended up seeing only 58-96 of the other 100, permanently. Not cosmetic — joinable? and mirrors? both read the roster, so you couldn't join someone you couldn't see, and a mate whose state was dropped wouldn't move you. The consumer now keeps the roster it is already relaying and hands a newcomer a snapshot in one message, so a join costs two broadcasts instead of N. Same 100 peers: roster complete and identical for everyone, 2310 messages sent down to 349, 72.6k deliveries down to 38.6k, server 30% -> 21% of one core, nav latency unchanged at ~14ms p50. This is a cache of relayed gossip, not a source of truth: parties are still worked out entirely in the browsers, and the server still decides nothing about them. It is process-local, like the in-memory channel layer it sits next to — if that ever moves to Redis for multiple workers, this moves too. Measured ceiling for the pathological case (everyone in ONE party, all mirroring each other): 100 peers ~14ms p50 at 21% of a core, 150 still ~14ms at 27%, 250 degrades to ~450ms p50 with the roster incomplete again. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> |
||
|---|---|---|
| scenes | ||
| server | ||
| tl | ||
| .dockerignore | ||
| .gitignore | ||
| DEPLOY.md | ||
| Dockerfile | ||
| fly.toml | ||
| manage.py | ||
| README.md | ||
| requirements.txt | ||
scene_analysis
tl/— the lumet frontend (ClojureScript / reagent / re-frame). Seetl/.- Django backend (this folder) — users, admin, and persistence of OTIO + the timeline scene.
Backend
The backend stores one Project per editing session: the uploaded OTIO
source file, the playback fps, and the live scene (the {:tracks :groups} mark-group map the frontend keeps in app-db) as a JSON dump.
Run
python -m venv .venv && .venv/bin/pip install -r requirements.txt
.venv/bin/python manage.py migrate
DJANGO_SUPERUSER_PASSWORD=admin .venv/bin/python manage.py createsuperuser --noinput --username admin --email admin@example.com
.venv/bin/python manage.py seed_demo # demo project from tl/.../one_two_three.otio
.venv/bin/python manage.py runserver 0.0.0.0:9001 # 0.0.0.0 so it's reachable over Tailscale/LAN
Admin: http://127.0.0.1:9001/admin/ (admin / admin). Upload OTIO and inspect the scene JSON there.
API (session auth; 401 if not logged in)
| Method | Path | Purpose |
|---|---|---|
| GET | /api/projects/ |
the current user's projects |
| GET | /api/projects/<id>/scene/ |
{fps, scene} |
| PUT | /api/projects/<id>/scene/ |
replace scene and/or fps (JSON body) |
| GET | /api/projects/<id>/otio/ |
serve the uploaded OTIO file |
| GET | /api/projects/<id>/revisions/ |
save history (who / when / summary) |
Attribution
Annotations are the authored unit, so that's what's attributed — and the server
is the authority (client-sent stamps are ignored, so authorship can't be
forged). On each scene PUT the server diffs incoming annotation groups against
the stored ones and stamps createdBy/editedBy (+ timestamps) from
request.user; it also writes a Revision (user, time, +N ~N −N summary,
snapshot of the annotation layer). A project is editable by its owner and any
collaborators (managed in the admin), so different users get distinct stamps.
The frontend currently loads /one_two_three.otio and persists annotations to
localStorage; pointing it at /api/projects/<id>/otio/ and the scene endpoints
is the next wiring step (CORS/credentials needed across the dev ports).