How to customize

It could well be that in your specific case, you need different steps to be taken to get to a deployable package. vdist by default is a bit naive: it checks for a requirements.txt and installs it using pip, and it also checks for a, on which it runs an install when present. Your situation might be a bit different. To solve this, vdist offers the ability to create custom build profiles. First, create a directory called buildprofiles under your project directory (location can be overridden by setting the profiles_dir argument on the Builder instance). In this directory, you place a script called profiles.json. The profiles.json file might look like this:

    "centos7": {
        "docker_image": "yourcompany/centos7:latest",
        "script": ""
    "debian": {
        "docker_image": "yourcompany/debian:latest",
        "script": ""

This configuration file defines 2 profiles: centos7 and debian. Each profile has 2 properties, called docker_image and script. The docker_image key indicates the name of the Docker image which will be pulled from the Docker registry by vdist. This can also be an image on your company's internal Docker registry, in which case the value for the docker_image property would look like "". The script key indicates what script to load on the build machine to actually execute the build process. These scripts are treated as Jinja2 templates, and the build information you provide to vdist will be injected into these templates.

By default, vdist provides the scripts and as generic templates for RHEL/CentOS/Fedora and Debian/Ubuntu based images, so you can use those when defining your profile. You can take a look at the templates provided by vdist to get an idea of how they work, and how to create your own. Custom shell scripts can be put in the buildprofiles directory alongside your profiles.json file. All parameters that are given by you when calling add_build() are injected into the template, including a few more. You can refer to these scripts in your own custom profiles, while using your own Docker images. For example: your company provides a provisioned build image based on Debian (custom Python interpreter package on board, regularly maintained and all), and refers to "" to perform the build.