Skip to content

dryewo/cyrus

Repository files navigation

A very opinionated Clojure project template

Build Status Clojars Project

Includes:

Usage

$ lein new cyrus org.example.footeam/bar-project +http +db +nrepl

Read below for the list of available options.

Additionally, you can use +all option that includes everything.

Contents

State management and dependency injection

mount-lite does great job managing start and stop order for application components:

;; Start everything in the correct order (first database, then webserver, etc.)
(m/start)

;; Stop everything in the reverse order
(m/stop)

Logging

Logging is supported via timbre.
Additionally, dovetail helper library provides useful tweaks:

  • Log level per namespace is configured in core.clj using log/set-ns-log-levels!. The resulting log level per namespace is the higher of global and specific settings.
  • All logging functions support throwable as their first argument.
  • All formatted values (%s) are automatically pr-stred.
  • Overall log level can be overridden by setting LOG_LEVEL environment variable.

HTTP server

+http option adds a component that starts an HTTP server (aleph). Example routes are provided as well as reasonable default middleware.

Swagger1st API

+swagger1st adds /api route that is handled by a separate library: Swagger1st.
It allows request routing and parameter parsing based on OpenAPI definition in YAML format. Highly recommended for any more or less serious API service.

Additionally exposes Swagger UI and spec:

  • /api/ui
  • /api/swagger.json

This option automatically includes +http.

A variant of this option +swagger1st-oauth2 includes OAuth2 protection of API endpoints, checking access tokens against Introspection Endpoint configured by TOKENINFO_URL environment variable. See fahrscheine-bitte for more information.

DB access

+db option adds a component that includes database access layer, built on:

  • HugSQL to generate access functions. Additionally, some interceptors are provided to convert camel_case column names to :kebab-case keywords in result maps.
  • conman for connection pool.
  • migratus for schema migrations. Migrations are always applied when the DB component starts up.
  • PostgreSQL JDBC driver is included.
  • Example schema is generated and example unit tests are provided.

PostgreSQL 9.6 for development and testing can be launched in a Docker container:

./make.sh db

Web UI

+ui option enables a very simplistic admin UI, served under /ui. HTML is rendered using hiccup. Static resources are served from resources/ui resource path, style.css is provided as an example.

A variant of this option +ui-oauth2 includes OAuth2 login with example configuration for GitHub. Authentication which can be configured to be optional or mandatory. See cyrus-ui-oauth2 for more information.

NREPL

+nrepl adds a NREPL server that is started before the main application is. It has to be enabled by setting NREPL_ENABLED=true. Default port is 55000, can be changed by setting NREPL_PORT.

Configuration management

The app can only be configured through environment variables (following Twelve-Factor App manifesto). Each environment variable has a corresponding cfg/def, which makes a checked and conformed value of that variable available as a global constant (var):

(cfg/def port "Port for HTTP server to listen on."
              {:var-name "HTTP_PORT"
               :spec     int?
               :default  8090})

On application startup all defined configuration constants are checked for validation errors and a summary is printed to console:

 INFO [main] m.core - Config loaded:
#'my.nrepl/bind: "0.0.0.0" from NREPL_BIND in :default // Network interface for NREPL server to bind to.
#'my.nrepl/port: 55000 from NREPL_PORT in :default // Port for NREPL server to listen on.
#'my.http/port: 8090 from HTTP_PORT in :default // Port for HTTP server to listen on.
#'my.db/password: <SECRET> because DB_PASSWORD is not set // Database password.
#'my.db/username: "postgres" from DB_USERNAME in :default // Database username.
#'my.db/jdbc-url: "jdbc:postgresql://localhost:5432/postgres" from DB_JDBC_URL in :default // Coordinates of the database. Should start with `jdbc:postgresql://`.
#'my.authenticator/tokeninfo-url: nil because TOKENINFO_URL is not set // URL to check access tokens against. If not set, tokens won't be checked.

For REPL-driven development a function is provided to override some configuration constants without restarting the REPL:

(cfg/reload-with-override! {"HTTP_PORT" 8888})

Please refer to the full documentation on cyrus-config home page.

user.clj

To facilitate REPL-driven development, user.clj contains functions, available directly from user namespace after REPL is started:

  • (start), which calls (m/start), first re-reading environment variable overrides from dev-env.edn.
    This is done to enable adjusting environment variables without restarting REPL.
  • (stop), which calls (m/stop) (and just in case reloads configuration too).
  • (reset), which calls (stop), (refresh) and then (start).
  • (tests), which runs all tests in all test namespaces.
  • functions from clojure.tools.namespace:
    • wrapped (refresh), which stops started states before reloading and then starts them again (so that you never lose references to running components).
    • (refresh-all), which reloads all the code in the project.

Development

In order to try the template out without releasing to clojars, install it to the local ~/.m2 and specify --snapshot flag:

$ lein install
$ cd target    # Or any other directory
$ lein new cyrus org.example.footeam/bar-project --snapshot -- +swagger1st +nrepl +db +http

Testing

$ ./itest.sh

FAQ

Q. Why another template? There is already Luminus, which is more feature-rich.
A. While Luminus is a great project, made with a lot of love, it already carries an opinion, which in some points is different from mine. Also, even without additional options it generates too much (not always you need front-end, configuration management with cprop is IMHO an overkill etc.)
Leiningen templates are not extensible, so I had to make my own, reusing best parts of Luminus.

Q. Why mount-lite and not component?
A. mount-lite is chosen over component for its lean spirit and ease of use. While in component you have to explicitly define dependencies between parts of the system, mount-lite (and mount) figure this out by scanning namespace declarations. This is a great example of DRY (don't repeat yourself) principle.
In mount-lite all states (components) have global names (like normal vars created with def) and are dereferable (implementing IDeref) — that means, you have to explicitly @ or deref to access them.

Q. Why mount-lite and not mount?
A. mount-lite is chosen over mount, because:

  • mount supports dereferable states via (in-cljc-mode), but when you try to deref a stopped state, mount automatically starts it. This leads to surprises and sometimes to impossibility to stop a running system (when you have a background process that accesses some state even after you execute (m/stop). The state just starts again, including the states it depends on).
    mount-lite in this case just throws an exception.
  • In mount it is only possible to start a part of the system (for testing it) by explicitly listing states that you need. In mount-lite there is "start-up-to" functionality that automatically figures out which states does this specific state need and starts only them (see examples in tests in a generated project).

License

Copyright © 2017 Dmitrii Balakhonskii

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.