It is still not possible to add a constraint to the URL route directly (in this example, the fact that the year should only consist of four digits). The current function django.conf.urls.url() will be replaced by django.urls.re_path() which will keep the existing functionality of allowing regular expressions to be placed in the definition of the route.

The functiondjango.conf.urls.include() will also be directly available under django.urls. This will allow you to import your modules like from django.urls import include, path.

Other New Things

• The Django Admin interface for the version 2.0 will be responsive, so it will be well adapted for use on mobile devices.
• A new Window class will be added, allowing for the use of the OVER keyword in SQL queries.
• A list of all new minor additions is available in the official documentation.

Helpful Links:

• Django 2.0 release candidate 1 released
• Django 2.0 release notes – UNDER DEVELOPMENT

Deploying Entire Services with Docker Swarm and GitLab CI

by Jacob Cook

Here at Savoir-faire Linux, I work on the deployment infrastructure for many of our hosted services, and am constantly looking for ways to improve our DevOps practice. Currently the deployment pipeline is centered around the use of Ansible to manually provision a service inside a virtual machine and adjust as necessary. But in an effort to make builds and deployments faster and more stable for our developers, we are looking at new approaches, including the use of Docker builds pushed to a registry and subsequently deployed using a pipeline like GitLab CI.

One of the tools that can be used to deploy a micro-service architecture (created with docker-compose) is called Docker Swarm. Using a set of servers configured to act in a swarm, one can deploy a stack of containers in a simple pattern with only docker stack deploy. Docker will then automatically choose the appropriate servers in the swarm to deploy to, and handle networking between the services just as if you were using docker-compose on your local machine. Neat, huh?

The best part is that it plays nicely with our existing build system in GitLab CI. We can build, test, release and deploy across our infrastructure using GitLab CI and GitLab Container Registry. There are still pieces of the puzzle to work out but it represents a major leap forward in the simplification and stability of our DevOps practice.

Further Reading:

• Deploy to Docker Swarm using GitLab CI
• Some tips for using Docker with Ansible