Well, Spark 3.5 still uses Parquet 1.13.x, so it can continue in that line,
we can always release a backported version if there is a security
issue. Spark 3.5 is their extended LTS so they will probably not move up to
a more recent version of Parquet. I imagine that the same applies for
Hadoop.

For HADOOP-19212 <https://issues.apache.org/jira/browse/HADOOP-19212> my PR
passes the tests with an upgrade 3.4.3.

Let's remember that we are talking about a new JDK requirement for the next
versions of Parquet, not for the previous ones.

So, what should we do? Do we need a vote?

On Wed, May 27, 2026 at 5:06 PM Aaron Niskode-Dossett via dev <
[email protected]> wrote:

> >
> >
> >    1. What does spark 3.5 require? Would a java17 requirement cause
> >    problems there?
> >
>
> Spark 3.5 supports versions 8, 11, and 17, so by dropping java 8 we've
> already diverged to some extent.
>
> https://spark.apache.org/docs/3.5.7/
>
> --
> Aaron Niskode-Dossett, Data Engineering -- Etsy
>

Reply via email to