[C++][Parquet] Support limit RowGroup size by bytes.

[zhf999]

Describe the enhancement requested

Could we support limit RowGroup size by bytes (before compression) in addition to RowGroup length (records count) to provide more precisely writing control.

Component(s)

C++

[Arbaaz123676]

Hi @zhf999 , I’d like to work on this issue.

Approach:

• Trace the existing C++ RowGroup sizing logic.
• Add an optional maximum uncompressed byte limit alongside the existing row-count limit.
• Preserve existing behavior when the byte limit is not configured.
• Add focused C++ tests for byte-based RowGroup boundaries and edge cases.

I’d appreciate it if you could assign this issue to me.

[zhf999]

Hi @zhf999 , I’d like to work on this issue.

Approach:

• Trace the existing C++ RowGroup sizing logic.
• Add an optional maximum uncompressed byte limit alongside the existing row-count limit.
• Preserve existing behavior when the byte limit is not configured.
• Add focused C++ tests for byte-based RowGroup boundaries and edge cases.

I’d appreciate it if you could assign this issue to me.

Hi @Arbaaz123676 !
I've created a PR to provide our solution, but it has not been merged yet.

#50743 #48468 These are previous works, you can also create a new PR if you are confident with your solution.