niquola / clunk

Clojure Postgres w/out JDBC

Clunk

/kləNGk/

noun

A heavy, dull sound such as that made by thick pieces of metal striking together.

Talking to Postgres via sockets. At this point in time, I'm just experimenting with Clojure; learning core.async.

Running The Tests

You need to have a Postgres (13) Database up and running.

docker compose up -d

Now you can run the tests with

clj -X:test

Example

(ns scratch.client
  (:require
   [com.clunk.core :as clunk]))

(def cfg {:username "jimmy"
          :password "banana"
          :database "world"})

We can now create a client, and connect to the postgres instance:

(def client (clunk/client cfg))

(clunk/connect client)

We will see a bunch of received messages:

{:type :AuthenticationMD5, :salt [-47 -110 29 -114]}
{:type :AuthenticationOk}
{:type :ParameterStatus, :param application_name, :status }
{:type :ParameterStatus, :param client_encoding, :status UTF8}
{:type :ParameterStatus, :param DateStyle, :status ISO, MDY}
{:type :ParameterStatus, :param integer_datetimes, :status on}
{:type :ParameterStatus, :param IntervalStyle, :status postgres}
{:type :ParameterStatus, :param is_superuser, :status on}
{:type :ParameterStatus, :param server_encoding, :status UTF8}
{:type :ParameterStatus, :param server_version, :status 13.4 (Debian 13.4-1.pgdg100+1)}
{:type :ParameterStatus, :param session_authorization, :status jimmy}
{:type :ParameterStatus, :param standard_conforming_strings, :status on}
{:type :ParameterStatus, :param TimeZone, :status Etc/UTC}
{:type :BackendKeyData, :id 167, :key 835881833}
{:type :ReadyForQuery, :status 73}

Finally we can run a query:

(clunk/query client "SELECT name FROM country LIMIT 1;")

Which prints out the following:

{:type :RowDescription, ...}
{:type :DataRow, ...}
{:type :Close, ...}
{:type :ReadyForQuery, ...}

And lastly, close the client:

(clunk/close client)

Rough Todo / Help needed

• Build out CLJS test suite
• Figure out a good way to provide async query results
• Fix core.async channel closing issues
• Flush out frontend/backend message parsers
• Deploy to Clojars
• Deploy to npm