Variable naming and reproducing is a bit confusing.

[arcoast]

I love Typetype but found the variables quite confusing initially.

For instance we define DATABASE_PASSWORD in the .env file but the stack will fall over unless we also manually propogate this change to postgres and postgres-init as well, and the confusing bit is the variables have different names in those containers.

I just wonder if it might be more intuitive to set the following variables in .env

# typetype-postgres
POSTGRES_DB=typetype
POSTGRES_USER=typetype
POSTGRES_PASSWORD=M5rzzxid6AFgG9ccE2ckpvY080Ai1R7L

and then use those in the compose file for DATABASE_USER and DATABASE_NAME and DATABASE_PASSWORD `

Then we can do the following with typetype-server

    typetype-server:
        image: ghcr.io/priveetee/typetype-server:latest
        container_name: typetype-server
        networks:
            typetype:
        # ports:
        #     - 8080:8080
        environment:
            ALLOWED_ORIGINS: ${ALLOWED_ORIGINS}
            DATABASE_URL: ${DATABASE_URL}
            DATABASE_USER: ${POSTGRES_USER}
            DATABASE_PASSWORD: ${POSTGRES_PASSWORD}

the following for typetype-downloader

    typetype-downloader:
        image: ghcr.io/priveetee/typetype-downloader:latest
        container_name: typetype-downloader
        networks:
            typetype:
        # ports:
        #     - 18093:18093
        environment:
            HTTP_PORT: ${HTTP_PORT}
            PUBLIC_BASE_URL: ${PUBLIC_BASE_URL}
            TYPETYPE_API_BASE: ${TYPETYPE_API_BASE}
            DB_URL: ${DB_URL}
            DB_USER: ${POSTGRES_USER}
            DB_PASSWORD: ${POSTGRES_PASSWORD}

and then for postgres

    postgres:
        image: postgres:17
        environment:
            POSTGRES_DB: ${POSTGRES_DB}
            POSTGRES_USER: ${POSTGRES_USER}
            POSTGRES_PASSWORD: ${POSTGRES_PASSWORD}
        volumes:
            - ${CONFIG}/typetype/typetype-postgres:/var/lib/postgresql/data
        restart: unless-stopped

and finally for postgres-init

    typetype-postgres-init:
        image: postgres:17
        depends_on:
            typetype-postgres:
                condition: service_started
        environment:
            POSTGRES_DB: ${POSTGRES_DB}
            POSTGRES_USER: ${POSTGRES_USER}
            PGPASSWORD: ${POSTGRES_PASSWORD}
        command:
            - /bin/sh
            - -ec
            - |
                until pg_isready -h typetype-postgres -U ${POSTGRES_USER} -d ${POSTGRES_DB}; do sleep 1; done
                EXISTS=$$(psql -h typetype-postgres -U ${POSTGRES_USER} -d ${POSTGRES_DB} -tAc "SELECT 1 FROM pg_database WHERE datname='typetype_downloader'")
                if [ "$$EXISTS" != "1" ]; then
                psql -h typetype-postgres -U ${POSTGRES_USER} -d ${POSTGRES_DB} -c "CREATE DATABASE typetype_downloader"
                fi
        restart: no

I've tested this and it works, and it means the end user only needs to change the variables in the .env file and it will be propogated correctly through the stack.

[arcoast]

Extrapolating this a bit more, if we add a variable

# typetype-postgres
POSTGRES_CONTAINER=postgres
POSTGRES_DB=typetype
POSTGRES_USER=typetype
POSTGRES_PASSWORD=Chill2-Gondola-Busload

then we can set the two DB_URL variables as a static value

# typetype-server
DATABASE_URL=jdbc:postgresql://${POSTGRES_CONTAINER}:5432/${POSTGRES_DB}

# typetype-downloader
DB_URL=postgres://${POSTGRES_USER}:${POSTGRES_PASSWORD}@${POSTGRES_CONTAINER}:5432/typetype_downloader?sslmode=disable

As long as the POSTGRES_* variables are at the start of the .env file the two database url variables will get derived automatically.

That also means the postgres-init-container can be set completely with variables

    postgres-init:
        image: postgres:17
        depends_on:
            typetype-postgres:
                condition: service_started
        environment:
            POSTGRES_CONTAINER: ${POSTGRES_CONTAINER}
            POSTGRES_DB: ${POSTGRES_DB}
            POSTGRES_USER: ${POSTGRES_USER}
            PGPASSWORD: ${POSTGRES_PASSWORD}
        command:
            - /bin/sh
            - -ec
            - |
                until pg_isready -h ${POSTGRES_CONTAINER} -U ${POSTGRES_USER} -d ${POSTGRES_DB}; do sleep 1; done
                EXISTS=$$(psql -h ${POSTGRES_CONTAINER} -U ${POSTGRES_USER} -d ${POSTGRES_DB} -tAc "SELECT 1 FROM pg_database WHERE datname='typetype_downloader'")
                if [ "$$EXISTS" != "1" ]; then
                psql -h ${POSTGRES_CONTAINER} -U ${POSTGRES_USER} -d ${POSTGRES_DB} -c "CREATE DATABASE typetype_downloader"
                fi
        restart: no

[Priveetee]

Hey, thanks for pointing this out.

Yeah I agree, this part is confusing right now.

The stack currently has DATABASE_* for the server, but Postgres itself still uses its own hardcoded values in a few places. So if someone changes the DB password in .env, they can easily end up with the server, downloader and Postgres not using the same value.

I think your idea makes sense: POSTGRES_DB, POSTGRES_USER and POSTGRES_PASSWORD should probably be the source of truth, and the server/downloader vars should reuse those.

So instead of having to update the same value in multiple places, users would only change it once in .env.

I'll clean this up in the compose files and will throw a PR to fix it. Thanks for catching it :p