GoDataDriven open source contribution: July 2018 edition

Welcome to the Open Source at GoDataDriven, July 2018 edition.

We start with Tünde and Kris who did a phenomenal job to add support to for Hive partitioned tables with partitions having different data formats in Spark. You can find the result of their work in PR 21893. Their work is the result of three sessions they had during our GoDataDriven Fridays. It involved determination, skills, and a bit of detective work throughout the Spark code base (they touched 7 files, adding more than 500 lines of code at the end).

The Spark folks are however reluctant to merge it. If you also think the feature is important and useful to you, let your voice be heard!

Vincent open sourced Asekuro, a tool to make it easier to test Jupyter notebooks, included notebooks that use the %load magic[^1.]

Fokko then directed his attention to the Event Hub-Spark connector by Microsoft, opening PR 356, 359, and 360.

Henk discovered another tool my Microsoft, DoWhy, a library that makes it easy to estimate causal effect. He immediately contributed PR 3 and 4.

Julian, on the other hand, also opened quite a large PR in Airflow, namely PR 3560. In there he pushes snakebite out and hdfs3 in to increase Airflow's compatibility with Python 3.

To close: I contributed PR 270 and 286 to dask-ml — although the first one might never be merged, even though it solves an open issue. Both PRs show a nice use of decorators, with the latest one also showing how to define context managers with yield.

That's it for this edition! Don't forget we're hiring! Especially if you are a software engineer that would like to move in the data space, get in touch as we're offering an apprenticeship starting from October.

And if you want more rambling throughout the month, follow me on Twitter: I'm gglanzani there!