One migration did a full-table UPDATE, a blocking GIN build and the trigger install together — long locks and a deploy that could outlast its health-check window on a large archive. Now two stages. 0002 only adds the nullable column and the trigger (metadata-only; new and updated rows get a vector from then on). 0003 is non-atomic: it backfills in 5,000-id batches, each its own short transaction and only touching rows still NULL, then builds the index with CREATE INDEX CONCURRENTLY (dropping a stale INVALID one from a failed earlier run first). Re-running is safe, and it's a cheap no-op on an install that applied the earlier single-step version. Measured on a real PostgreSQL 16 with 150,000 existing articles while a second connection kept inserting/updating: the migration took ~20s, the slowest concurrent write was 37ms (nothing blocked), no NULL vectors remained, the index was valid, and unapply/re-apply worked. Full suite passes on Postgres. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Evection
A self-hosted RSS/Atom feed reader, in the spirit of FreshRSS, built in Python (Django) instead of PHP. Installable as a PWA on your phone, with offline reading for anything you've already loaded.
Features
- Multi-user, with per-user folders, read/starred/saved-for-later state, and tags
- Background feed polling via Celery, with conditional GET and backoff on errors
- Full-text search over your articles (SQLite FTS5 or PostgreSQL, your choice)
- A filter/rules engine to auto-tag, star, hide, or mark feeds' articles read
- OPML import/export
- Installable as a PWA (Add to Home Screen) with an offline fallback page
- An admin dashboard surfacing feeds that are failing to fetch
Deploying your own instance
See docs/install.md for a production install: pip + gunicorn
- systemd + nginx, with both SQLite and PostgreSQL covered.
Development setup
python3 -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
cp .env.example .env
python manage.py migrate
python manage.py runserver
Visit http://localhost:8000/ — you'll land on the first-run setup wizard to
create the instance's admin account.
Updating pinned dependencies
requirements.txt/requirements-dev.txt are hash-pinned lock files generated
from pyproject.toml — production installs (scripts/install.sh,
scripts/upgrade.sh) install from requirements.txt with --require-hashes
rather than re-resolving version ranges at install time. After changing a
dependency in pyproject.toml, regenerate both and commit them alongside:
pip-compile --generate-hashes --allow-unsafe --output-file=requirements.txt pyproject.toml
pip-compile --generate-hashes --allow-unsafe --extra=dev --output-file=requirements-dev.txt pyproject.toml
Tests
pytest
ruff check .
black --check .
Testing against PostgreSQL
The suite runs on SQLite by default; the Postgres-only tests (search backend) are skipped there. To run everything against a real Postgres:
docker run -d --name evection-pg-test -e POSTGRES_PASSWORD=testpw -e POSTGRES_DB=evection \
-p 127.0.0.1:55432:5432 postgres:16
DATABASE_URL=postgres://postgres:testpw@127.0.0.1:55432/evection pytest
docker rm -f evection-pg-test
License
GNU AGPL-3.0-or-later — if you run a modified version of Evection as a network service, you must make your modified source available to its users.