about a panic in rle decoder

[jp0317]

The rle decoder has an assertion that can cause panic given corrupted inputs and the corresponding unit test says TBD (currently panics). By TBD It seems the panic is just a temporary behavior. If so, what would be the better behavior there if we want to avoid panics?

[jp0317]

Hi @alamb @tustvold, do you think this is a case where panic should be error results? Or should the reload() just return false (similar to arrow-cpp behavior) if failed to get current_value?

[alamb]

Hi @alamb @tustvold, do you think this is a case where panic should be error results? Or should the reload() just return false (similar to arrow-cpp behavior) if failed to get current_value?

If the input is malformed I think the Rust parquet reader should return an error , following the guidelines in https://github.com/apache/arrow-rs?tab=readme-ov-file#guidelines-for-panic-vs-result

[Jefffrey]

Raised #7102 as this seems like a bug according to the guidelines in the README