Table of Content

• 1. Chat Server with Mojo
• 2. Chat Server with Redis
• 3. Chat Server with PostgreSQL
• 4. Chat Server with PAGI
• 5. Chat Server with Thunderhorse

---

1. Chat Server with Mojo

Start chat server as below:

Source

$ perl chat-server.pl daemon

2. Chat Server with Redis

I was introduced to JSON Virtual Frame and JSON Virtual Event when I shared the previous version of chat server.

To be honest, I had no clue about any of them, so I dig deep and found this:

JSON Virtual Frame

This is a special way to structure data when you send a message through a WebSocket. Instead of manually converting a Perl data structure into a JSON string, you use the json key in the hash you pass to the send() method.

How it works?

When you call $tx->send({json => $data}), the build_message method in Mojo::Transaction::WebSocket intercepts this. It automatically calls Mojo::JSON::encode_json to serialise your Perl $data into a JSON text string. It then sends this string as a standard WebSocket text frame.

Why use it?

It makes your sending code cleaner and less error-prone by removing the manual encode_json step.

Example:

# Instead of doing this manually:
use Mojo::JSON 'encode_json';
$tx->send({text => encode_json({message => 'Hello', user => 123})});

# You can do this directly:
$tx->send({json => {message => 'Hello', user => 123}});

JSON Virtual Event

This is a special event you can listen for on the WebSocket transaction. When a complete WebSocket message arrives, if the json event has any subscribers (i.e., you have an on(json => ...) callback), the transaction will automatically attempt to decode the message payload from JSON.

How it works?

When a message is fully assembled (in parse_message), it checks if there are any subscribers for the json event. If there are, it passes the raw message through Mojo::JSON::j() (which is context-aware and decodes JSON). The decoded Perl data structure is then emitted to your callback.

Note: This event is only emitted if you are listening for it. This ensures you don't incur the performance cost of decoding JSON if you don't need it. It can decode both text and binary messages, as long as they contain valid JSON.

Example:

$tx->on(json => sub {
  my ($tx, $received_data) = @_;
  say "Received message: " . $received_data->{message};
});

I have applied both in this version.

For this version of chat server, we need Redis running locally. Luckily I had docker container running Valkey.

If you don't know Valkey, then I suggest you take a look at this blog post of mine.

First start the Valkey container as below:

$ docker start valkey

Testing time, make sure it reachable on default port as below:

$ redis-cli -h 127.0.0.1 -p 6379 ping

You should see PONG in response to the above command.

Time to start the chat server listening to port 3000 like below:

Source

$ perl chat-server-v2.pl daemon -l http://*:3000

Let's start another chat server this time listening to port 3001 like below:

$ perl chat-server-v2.pl daemon -l http://*:3001

Now open browsers to both ports, users on different ports can now chat to each other.

3. Chat Server with PostgreSQL

Well, we need PostgreSQL database now and I am not willing to setup database from scratch.

The easy option is to create a docker container running the PostgreSQL database.

So here is the docker compose configuration file: docker-compose.yml

version: '3.8'

services:
  postgres:
    image: postgres:16-alpine
    container_name: chat_postgres
    restart: unless-stopped
    environment:
      POSTGRES_USER: chatuser
      POSTGRES_PASSWORD: chatpass
      POSTGRES_DB: chat_db
    ports:
      - "5432:5432"
    volumes:
      - postgres_data:/var/lib/postgresql/data
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U chatuser -d chat_db"]
      interval: 10s
      timeout: 5s
      retries: 5

volumes:
  postgres_data:

Start the container first as below:

$ docker-compose up -d

Check the status of the container as below:

$ docker-compose ps

Test the connection now as below:

$ docker exec -it chat_postgres psql -U chatuser -d chat_db
psql (16.11)
Type "help" for help.

chat_db=# LISTEN chat_messages;
LISTEN
chat_db=# NOTIFY chat_messages, '{"type":"test"}';
NOTIFY
Asynchronous notification "chat_messages" with payload "{"type":"test"}" received from server process with PID 2333.
chat_db=# \q

We have PostgreSQL up and running.

Let's start two chat servers listening to port 3000 and 3001.

Source

$ perl chat-server-v3.pl daemon -l http://*:3000

$ perl chat-server-v3.pl daemon -l http://*:3001

Once again, open browsers to both ports, users on different ports can now chat to each other.

4. Chat Server with PAGI

Start chat server as below:

Source

$ perl chat-server-v4.pl

5. Chat Server with Thunderhorse

To run this server, we need Perl v5.40 as enforced by Thunderhorse,

You will also need PAGI.

Source

$ pagi-server --port 3000 chat-server-v5.pl