Possible bug in parquet-mr

[Sky Brewer <sk...@ess.eu.INVALID>]

Hi all,

I found a possible bug in parquet-mr. I found if you set up a read over a parquet file with a filter that ends at the end of the file. Then read the file again with a new filter, the last event read will be updated to be the last event in the file, regardless of the file.

I’ve set up a reproduction in https://github.com/jacomago/parquetentrybug where the test src/test/java/MainTest.java<https://github.com/jacomago/parquetentrybug%20where%20the%20test%20src/test/java/MainTest.java>. Should pass, but currently does not.

Thanks
Sky

Re: Possible bug in parquet-mr

[Gang Wu <us...@gmail.com>]

Hi Sky,

Thanks for reporting the issue! Could you please open a PR to parquet-mr
with a minimal reproducible test case? That would be a lot easier for the
investigation.

Best,
Gang


On Tue, Jan 23, 2024 at 4:23 PM Sky Brewer <sk...@ess.eu.invalid>
wrote:

> Hi all,
>
> I found a possible bug in parquet-mr. I found if you set up a read over a
> parquet file with a filter that ends at the end of the file. Then read the
> file again with a new filter, the last event read will be updated to be the
> last event in the file, regardless of the file.
>
> I’ve set up a reproduction in https://github.com/jacomago/parquetentrybug
> where the test src/test/java/MainTest.java<
> https://github.com/jacomago/parquetentrybug%20where%20the%20test%20src/test/java/MainTest.java>.
> Should pass, but currently does not.
>
> Thanks
> Sky
>
>
>