[C++][CI] Use differential fuzzing on IPC file fuzzer

[pitrou]

Describe the enhancement requested

The IPC file fuzzer could also try to read the file as an IPC stream and validate that it reads the same.

Component(s)

C++, Continuous Integration

[vanshaj2023]

Hi! I'd like to work on this issue.

Implementation Approach:

1. Modify the IPC file fuzzer to write the generated data as both IPC file and IPC stream formats
2. Read back both formats and compare the resulting RecordBatches/Tables to ensure they contain identical data
3. Add validation logic to check schema, metadata, and actual data values match between both reads
4. Handle edge cases where format-specific features might cause legitimate differences

This would help catch bugs where the file and stream readers behave inconsistently on the same data.

Could you please assign this issue to me? Also, does this approach sound correct, or is there anything I'm missing or should do differently?

Thanks!

[pitrou]

@vanshaj2023 1) There is already a PR open for this 2) Please don't post random AI-generated messages on the issue tracker. Thanks in advance!

[vanshaj2023]

Hi @pitrou

My bad! I should have checked for existing PRs first, and I apologize for the AI-generated comment. I'm still learning how open source contributions work.

Thanks for pointing that out!

[pitrou]

Issue resolved by pull request 49312
#49312

[pitrou]

@vanshaj2023 You're welcome. Also, there are issues marked with the labels "good first issue" or "good second issue" for people without much experience contributing to Arrow.