Release DataFusion `50.3.0` (minor)

[hareshkh]

Is your feature request related to a problem or challenge?

• Follow on to Release DataFusion 50.2.0 (September 2025) Release DataFusion 50.2.0 (minor) #17849

There are some things that we may way to backport into the 50.0.0 line, and this ticket tracks doing so.

Steps

• Update version and changelog
• Test with DataFusion Python
• Test with DataFusion Comet Release DataFusion 50.3.0 (minor) #18072
• Test with delta.rs:
• Test Vortex
• Test with iceberg-rust:
• Test with LakeSail:
• Test with parquet viewer
• Test with datafusion-materialized-views
• Voting Thread
• Publish to crates.io
• Create ticket for next release:

Potential Content:

Minor Features

• feat: support Spark concat string function #18063 + chore: use NullBuffer::union for Spark concat #18087 @comphead

Bug Fixes

• AmbiguousReference on project following join when expression name matches left or right column name #17294 / Fix ambiguous column names in substrait conversion as a result of literals having the same name during conversion. #17299 @xbailey
• LimitPushPastWindows returns incorrect results for queries with lead() #18028 / Fix bug in LimitPushPastWindows #18029 @avantgardnerio
• Semantic equality check for DataType::Timestamp doesn't exist #17776 / Extend datatype semantic equality check to include timestamps #17777 @shivbhatia10
• change_redundant_column lossy approach breaks logical optimizer and physical planner #17405 / fix(SubqueryAlias): use maybe_project_redundant_column #17478 @notfilippo
• DateAdd / DateSub do not perform overflow checks in release builds datafusion-comet#2539 / fix: Add overflow checks to SparkDateAdd/Sub to avoid panics #18013 @andygrove
• Datafusion 50 Performance Regression (array_has style filter/join for Parquet data set) #18070

Backport PR list:

• [branch-50]: chore: cherry pick concat to 50.3.0 #18128
• [branch-50] Fix ambiguous column names in substrait conversion #17299 #18077
• [branch-50] Backport Fix bug in LimitPushPastWindows (#18029) #18107
• [branch-50] Extend datatype semantic equality check to include timestamps (#17777) #18129
• [branch-50] fix(SubqueryAlias): use maybe_project_redundant_column (#17478) #18130
• [branch-50]: fix: Add overflow checks to SparkDateAdd/Sub to avoid panics (#18013) #18131

[alamb]

FYI @xudong963 -- I can do this one too, but wanted to make sure you were aware

Thank you @hareshkh for filing the issue

[comphead]

Thanks folks

• perhaps feat: support Spark concat string function #18063
  Please allow me couple of hours to test it out

[alamb]

Another potential candidate:

• Datafusion 50 Performance Regression (array_has style filter/join for Parquet data set) #18070

[andygrove]

#18013 would be nice to have

[avantgardnerio]

I created #18107 . Sorry for the delay!

[xudong963]

@alamb, I can be your backup in case you have something urgent at that time.