delta-rs VS chispa

I work at Databricks, but am pretty must just an OSS nerd, mainly focusing on Delta Rust recently: https://github.com/delta-io/delta-rs

I did some keyword research and wrote this post cause lots of folks are doing searches for Delta Lake vs Parquet. I'm just trying to share a fair summary of the tradeoffs with folks who are doing this search. It's a popular post and that's why I figured I would share it here.

Seeing a lot of great libraries coming out with python bindings in the data world e.g delta-rs Polars. I see it growing in this space as a C++ alternative

If I’m not misunderstanding, you could both decode the gRPC protobuf AND write to delta lake in Rust. Tonic, Delta-rs.

https://github.com/delta-io/delta-rs/issues/850 … looks like it’s on the roadmap!

Microsoft talk a lot about OneLake and that the delta file format will be the standard during the build conference. Is it only me that find it strange that their marketing team talks so much about the delta format when they do not even provide a library to work with the delta format from .net? It would be easy for them to maintain bindings to https://github.com/delta-io/delta-rs but also provide a reader that support V-Order https://learn.microsoft.com/en-us/fabric/data-engineering/delta-optimization-and-v-order?tabs=sparksql

I know someone will be adding this on the python side in the coming weeks. On the rust side you can use delta-rs with polars. Though you would be compiling both arrow2 and arrow-rs, so that's quite heavy.

You don’t need DBX to use Delta Lake. You can use S3 as the backend and just use the Python Delta Lake library. It works great! https://github.com/delta-io/delta-rs

Maybe if I get some free time soon I can formalize into a working example. Been wanting an excuse to try similar concept in delta-rs and polars/duckdb vs databricks/spark vs iceberg/polars.

delta-rs is a native Rust library for Delta Lake. It's a better way to store data than Parquet files and is fundamentally important library for the Rust data ecosystem. It's tightly integrated with Polars and Datafusion and there is a lot of interesting Rust work to be done.

Unit and e2e tests using a combination of pytest and chispa (https://github.com/MrPowers/chispa). Custom library to create random test data that fits schema with optional hardcoded overrides for relevant fields to test business logic.

here's a little README fix a user pushed to chispa

chispa is a library of PySpark testing functions.

The other approach I've used is Poetry, see the chispa project as an example. Poetry is especially nice for projects that you'd like to publish to PyPi because those commands are built-in.

- All Spark transformations are tested with pytest + chispa (https://github.com/MrPowers/chispa)

Pandas syntax is far inferior to regular PySpark in my opinion. Goes to show how much data analysts value a syntax that they're already familiar with. Pandas syntax makes it harder to reason about queries, abstract DataFrame transformations, etc. I've authored some popular PySpark libraries like quinn and chispa and am not excited to add Pandas syntax support, haha.

I am the author of spark-fast-tests and chispa, libraries for unit testing Scala Spark / PySpark code.

Blogging has been the main way I've been able to attract users. Someone searches "testing PySpark", they see this blog, and then they're motivated to try chispa.

I built daria (https://github.com/MrPowers/spark-daria) to make it easier to write Spark and spark-fast-tests (https://github.com/MrPowers/spark-fast-tests) to provide a good testing workflow.

quinn (https://github.com/MrPowers/quinn) and chispa (https://github.com/MrPowers/chispa) are the PySpark equivalents.

Built bebe (https://github.com/MrPowers/bebe) to expose the Spark Catalyst expressions that aren't exposed to the Scala / Python APIs.

Also build spark-sbt.g8 to create a Spark project with a single command: https://github.com/MrPowers/spark-sbt.g8

I've built popular PySpark (quinn, chispa) and Scala Spark (spark-daria, spark-fast-tests) libraries.