If you are seeing this, you are one of the very first people to see Zero outside of Rocicorp. That must mean we think a lot of you!

This is still early. There are still many bugs. Basically you can run this dogfood app, get a feel for what Zero will be like, and tinker with some queries. You won't be able to write your own app. But we still think it's pretty encouraging in its fledgling form.

We do not yet have any npm packages – Zero is under rapid development and we're building it side-by-side with this demo app. The best way to play with Zero is to just play with the demo app.

From root of monorepo:

npm install

cd apps/zeppliear/docker
docker compose up

This will take some time to populate the database with test data the first time.

Create a .env file in the zeppliear directory:

# The "upstream" authoritative postgres database
# In the future we will support other types of upstreams besides PG
UPSTREAM_URI = "postgresql://user:password@127.0.0.1:6432/postgres"

# A separate postgres database we use to store CVRs. CVRs (client view records)
# keep track of which clients have which data. This is how we know what diff to
# send on reconnect. It can be same database as above, but it makes most sense
# for it to be a separate "database" in the same postgres "cluster".
CVR_DB_URI = "postgresql://user:password@127.0.0.1:6433/postgres"

# A separate postgres database we use to temporarily store replication messages.
# This facilitates running multiple replicas or catching up replicas from
# asynchronous backups without the expense of using multiple upstream Postgres
# replication slots. It can be same database as above, but it makes most sense
# for it to be a separate "database" in the same postgres "cluster".
CHANGE_DB_URI = "postgresql://user:password@127.0.0.1:6433/postgres"

# Uniquely identifies a single instance of the zero-cache service.
REPLICA_ID = "r1"

# Place to store the SQLite data zero-cache maintains. This can be lost, but if
# it is, zero-cache will have to re-replicate next time it starts up.
REPLICA_DB_FILE = "/tmp/sync-replica.db"

# Logging level for zero-cache service.
LOG_LEVEL = "debug"

Then start the server:

npm run start-zero-cache

This will take some time to populate the replica.

In still another tab:

VITE_PUBLIC_SERVER="http://127.0.0.1:3000" npm run dev

After you have visited the local website and the sync / replica tables have populated.

rm -rf /tmp/sync-replica.db

docker compose down
docker volume rm -f docker_pgdata_upstream

• Sync seems solid-ish, but we've only been running it for a week. If you see it fail, you can recover by clearing your local browser state and refreshing.
• Filter by label doesn't work at all.
• Mutation propagation is slow if there are priority or status filters selected.
• All tables synced with Zero require a id primary key, even junction tables. This is not a requirement of our query engine it's just left over goop we haven't fixed yet.