Codecov Report

Base: 83.12% // Head: 83.12% // Increases project coverage by +0.00% 🎉

Coverage data is based on head (a22cce5) compared to base (1fcfd7c).
Patch coverage: 84.03% of modified lines in pull request are covered.

❗ Current head a22cce5 differs from pull request most recent head bbd2b3a. Consider uploading reports for the commit bbd2b3a to get more accurate results

@@           Coverage Diff           @@
##             main    #1326   +/-   ##
=======================================
  Coverage   83.12%   83.12%           
=======================================
  Files         370      370           
  Lines       40169    40241   +72     
=======================================
+ Hits        33391    33451   +60     
- Misses       6778     6790   +12     

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

O(n) is restored:

@jorgecarleitao I added a test list_nested_inner_required_required_i64 which successfully round trips, only the statistics are not equal to pyarrow.

I am not really sure what the proper statistics should be in such a case. The values are correct only the nesting differs. Pyarrow does not nest the null count, whilst we do.

---- io::parquet::write::list_nested_inner_required_required_i64 stdout ----
thread 'io::parquet::write::list_nested_inner_required_required_i64' panicked at 'assertion failed: `(left == right)`
  left: `Statistics { null_count: UInt64[0], distinct_count: UInt64[None], min_value: ListArray[[[0]]], max_value: ListArray[[[10]]] }`,
 right: `Statistics { null_count: ListArray[[[0]]], distinct_count: ListArray[[[None]]], min_value: ListArray[[[0]]], max_value: ListArray[[[10]]] }`', tests/it/io/parquet/write.rs:59:5
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

Oof, thank you so much, @ritchie46 !

@jorgecarleitao This should be good to go. The coverage failure is unrelated.