Getting Started

To get started with praxxis, you need

To install in production mode, move to the root of the repo in the command line and run

pip install .

This will install praxxis onto your machine, and allow you to run commands by typing

prax

On your first run, praxxis will set up the directories and databases it needs to store data and run.

Importing Libraries

praxxis can import libraries from anywhere on your machine or from a hosted git service. Right now, the project supports GitHub, GitLab, and BitBucket for remote repositories.

To get started with test notebooks, navigate to the root of the project and run

prax al tests/test_notebooks

This will load the pre-written test notebooks into your session.

To check that the library has been loaded properly, run

prax ll

And verify that the library “test_notebooks” is loaded.

To check that the notebooks have been loaded properly, run

prax l

You should be able to see the notebooks “broken_notebook,” “test_notebook,” and “test_param_inject”.

You can also import a library of notebooks from a git remote by using the same al command.

prax al <your git url>

Using Notebooks

Now that you have your notebooks loaded into praxxis, you can execute them!

First view what notebooks you have loaded by running

prax l

This should display the notebooks and the libraries they are located in.

Running notebooks

In order to run a notebook, run

prax <ordinal or name>

praxxis also supports running a notebook with the html flag, which opens the output file in a browser when execution completes. This will also save an HTML version of the output to your machine.

prax <ordinal or name> html

Opening notebooks

To open a notebook, praxxis offers a few options.

prax o <name or ordinal> html
prax o <name or ordinal> jupyter
prax o <name or ordinal> ads

opens notebooks in HTML, Jupyter, or Azure Data Studio, respectively. To open the notebook in vim, use the default command

prax o <name or ordinal>

Using Parameters

Some notebooks have parameters that can be injected, allowing the notebook’s functionality to be changed without editing the notebook itself. Praxxis supports parameter injection into any notebook with the first code cell tagged as a parameter cell.

Notebook Parameters

To view the parameters that can be set in a certain notebook, run

prax v <notebook name or ordinal>

to pull those parameters out of a notebook and into your scene, run

prax p <notebook name or ordinal>

Library Parameters

to view the parameters that can be set throughout a library, run

prax vl <library name or ordinal>

to pull those parameters out of the library and into your scene, run

prax pl <library name or ordinal>

Parameter Operations

Existing parameters can be set or updated by running

prax sp <name or ordinal> <value>

deleted by running

prax dp <name or ordinal>

and viewed by running

prax lp

To add a new parameter, set a parameter with the desired name and value:

prax sp <name> <value>

To try out injecting a parameter yourself, run these commands in the root of your praxxis folder.

prax al tests/test_notebooks
prax l
prax v test_param_inject
prax p test_param_inject
prax sp text_to_print 'hello test!'
prax r test_param_inject

This sequence:

Using Scenes

Scenes allow you to create environments to keep your work in one place. Scenes save parameter configurations and execution history. You can indicate whether a problem has been resolved for now by ending and resuming scenes.

To make a new scene, run

prax ns <name of new scene>

To list scenes, run

prax ls

Scenes can be marked as ended to signify that a workflow has finished. To end a scene, run

prax es <name or ordinal>

Scenes can also be resumed with the command

prax rs <name or ordinal>

To delete a scene, removing the associated history and parameter values from your machine, run

prax ds <name or ordinal>

Setting Up Telemetry

By default, praxxis stores no information on your notebook habits. However, to enable nightly improvements on the machine learning model used for some predictions, you need to link your installation to some type of external storage. This allows for a machine learning model to be trained on your habits as well as the habits of every other praxxis user using the same storage pool.

In order to access the telemetry setup utility, run

prax u

and use ordinals to select settings.

Telemetry

to turn telemetry on and off, select telemetry from the list. Set the value to 1 for on, and 0 for off.

URL

this is the url of the storage pool. The default url is set to the configuration for a 2019 SQL Server Big Data Clusters instance, but it can be changed to anything.

Host

this is the IP of the host machine that you want to send data to.

Username

this is the username of the login that you want to use to send data.

Password

this is the password of the login that you want to use to send data.

If you’ve configured these settings correctly, then you should be able to use praxxis and see data on notebook execution appearing in your cluster. If these settings are not configured correctly, you will see warnings after each notebook execution about the backlog of unsent file outputs.