Complete pproject workflow

The following figure illustrates how a typical workflow inside pproject usage looks like and which stations in development-circle correspond to which pproject-command.


Using groups inside vcs


Groups currently can only be used when using gitlab as vcs inside config. pproject doesn’t support github-teams yet.

Valid group formats:

Using groups create the possibility to set different visibility rules and access authorizations for each kind of project individually.

Developing inside conda-environments

Projects often depend on a specific python-version and multiple third-party packages like psycopg2, attrs, pandas, etc. The syntax and functionalities of these packages may change or be deprecated in former or future versions. Thats why it is always important to develop inside strict defined environments. There are a lot of possibilities for that: using pip with virtualenv, pipenv, docker, etc. One of these possibilities are conda-environments which are used in pproject.

Using the meta.yaml as conda-environment definition

The idea to use the meta.yaml as the environment-definition instead of using the use environment.yml file was inspired by David Wölfle.

At this point:
Thanks a lot to David for this awesome idea!

If you build a conda-package, the meta.yaml definition is used to collect the package dependencies. Using an additional environment.yml is error-prone and redundant information. Creating an environment based on the information in the meta.yaml at the requirements-run-section allows more flexibility for the versions of the dependencies (you can set >, <, >=, <=, and x* version-numbers) then in the environment.yml definition which uses exact version-definitions. You also make sure to develop in the same environment as resulting in the environment in which you install the resulting package. It also provide the development of a project with multiple developers, all using the same environment like they would use the same environment.yml-file.

The pproject update and pproject autoenv commands collect the dependency-informations as defined in the meta.yaml’s requirements-run-section and creates an conda-environment based on these dependencies. If the environment already exists, it will be removed and recreated based on the new definiton. This pretends the environment to include dependency-zombies from former environment-definitions. So make sure to place all your dependencies inside the requirements-run-section of the meta.yaml.


All other installations (conda install) and deletions (conda remove) inside the conda environment will be reset after the next pproject update execution (either triggered by pproject autoenv or manually run).

If pproject autoenv recognizes a change inside the meta.yaml file (based on md5sum-comparison saved in the hash.md5 file), pproject update is triggered. If you didn’t activate the autoupdate with pproject autoupdate_toggle (in .bashrc/.zshrc or manually), you can run the pproject update command yourself.

Using entry-points

If you developed a products, operations or services package it is meaningful to set an entry point for your main function inside the meta.yaml file. This allows to run your project after build with a short command like

/var/local/conda/envs/ouroboros-products-dummy/bin/dummy --help

You can also use:

source activate ouroboros-products-dummy
dummy --help

If you don’t use an entry-point you’ll have to use a possibly really long command like the following to execute your project:

/var/local/conda/envs/ouroboros-products-dummy/bin/python /var/local/conda/envs/ouroboros-products-dummy/lib/python3.6/site-packages/ouroboros-products-dummy/ouroboros/products/dummy/ --help

Thats why it is recommended to use entry-points. Especially if you want to use readable cronjos for your project or if your resulting project is used often.

pproject project structure

The used skeletons can be found at:

├── build (after using sphinx)
├── conda-build
│   ├── hash.md5
│   └── meta.yaml
│   ├──
│       ├── PROJECT
│       │   ├──
│       │   └──
│       └──
├── source (after using sphinx)
└── tests

Your project dependencies are stored in conda-build/meta.yaml. Whenever you add a new dependency (import) to your code, add this dependency to the requirements-run-section inside your meta.yaml. If autoenv is set inside the precommand and autoupdate_toggle is on, pproject detects the changes in your meta.yaml file and automatically rebuilds your conda-environment. If autoenv is disabled you can run this step manually from inside the folder company-namespace-project with

pproject update

If autoenv is activated the corresponding conda-environment for your project is activated in shell when entering the company-namespace-project-folder. If you leave this folder, the environment is deactivated. Please note, that any dependency added to your environment with “conda install” is removed from the environment if pproject update is triggered (either from autoenv or manually). All dependencies have to be placed inside the meta.yaml. Important: don’t edit the hash.md5 file, the content of this file is managed by pproject to decide if rebuild of the environment is necessary or not.


Please store all your tests in the tests-folder, so pproject test can find them. Don’t edit the if not really needed and you’re not really sure what you are doing (the content is automatically build from the pproject create command and is required for the conda-package build process).

© Copyright 2018, Simon Kallfass.


This site was generated using Sphinx and the Sphinx-Bootstrap-Theme