search

awslabs / python-deequ

Python API for Deequ
Apache License 2.0
713 stars 134 forks source link

Upgrade to deequ 2.0.4-spark-3.3 dependency #196

Closed rdsharma26 closed 5 months ago

rdsharma26 commented 5 months ago

Description of changes:

This commit updates the Spark 3.3 dependency of Deequ. There are some breaking changes to the Scala APIs, from a Py4J perspective. In order to work around that, we use the Spark version to switch between the updated API and the old API. This is not sustainable and will be revisited in a future PR, or via a different release mechanism. The issue is that we have multiple branches for multiple Spark versions in Deequ, but only one branch in PyDeequ.

The changes were verified by running the tests in Docker against Spark version 3.3. The docker file was also updated so that it copies over the pyproject.toml file and installs dependencies in a separate layer, before the code is copied. This allows for fast iteration of the code, without the need to install dependencies every time the docker image is built.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

chenliu0831 commented 5 months ago

Nice - this seems to be a short term solution for https://github.com/awslabs/python-deequ/issues/169.

Also would this address the issue we run into for 3.4+ https://github.com/awslabs/python-deequ/pull/168?

rdsharma26 commented 5 months ago

@chenliu0831 #168 will likely not be fixed by this. I'll address the upgrade of Spark 3.4 and the associated changes in a separate PR.

chenliu0831 commented 5 months ago

@rdsharma26 got it - I think the title of the PR might be more accurate if it's about "Deequ 2.0.4" parity