Hi everybody,
thanks for the many opinions!
>> I'm leaning in favor of option 1 and try to release Nutch 1.22
>> during the next weeks. After the release go to option 3.
>
> I think this is a good compromise. We could even spin 1.23 fairly soon
> afterwards with scope limited strictly to JDK17 updates.
Let's do this.
After scanning over the Hadoop repositories and mailing lists:
- The trunk build now uses JDK 17. The switch from JDK 8 was
very recent [1].
- The release of Hadoop 3.5.0 (on JDK 17) is on the way [2]
with some work to be done.
This will allow us to move to Java 17 and also to stop
trying to be backward-compatible with further releases.
~Sebastian
[1]
https://github.com/apache/hadoop/commit/c8b0df33e903b34595153d87faaded6b0024dd67
[2] https://lists.apache.org/thread/y7jgg1k60rcn2xddw59b8rprk403ssxf
https://lists.apache.org/thread/0lnrbxlfdb0clq8h8rfpbjfftc53db6o
On 2/6/26 00:08, Lewis John McGibbney wrote:
Thanks for kicking this thread off Seb.
A while back I also registered https://issues.apache.org/jira/browse/NUTCH-3045
which details other considerations for upgrade to JDK17. Although these are not
100% necessary, I would argue they are absolutely features of JDK17 we should
seriously consider incorporating into Nutch.
On 2026/02/03 21:52:04 Sebastian Nagel wrote:
the current Nutch development is ready for Java 17
with NUTCH-2971 fixed - thanks Isabelle!
Yes thanks Isabelle!
We now have the following options for the next release (1.22):
Thank you for laying out some options.
1. stay on Java 11
RE #1: To me this presents the best risk averse option but my least favorable.
I am not keen staying with JDK11 due to EOL. I would like to upgrade to JDK17.
We could always roll a 1.21.1 with JDK11 support if we discover moving to JDK17
breaks lots of deployments.
2. require Java 17 at compile time, but compile using "-target 11"
to stay compatible with Java 11 at runtime
RE #2: I don't see much value in this approach over either 1 or 3.
3. drop support for Java 11 and switch to Java 17
RE #3: This is my preferred approach. It addresses the EOL issue, (should
hopefully) unblock further development opportunities and allows us to move with
the times. I understand the foundational JDK11 requirements levied by Hadoop
3.X. We need to smoke/integration test with Nutch master branch with JDK17 and
ensure interoperability with Hadoop JDK11.
I'm leaning in favor of option 1 and try to release Nutch 1.22
during the next weeks. After the release go to option 3.
I think this is a good compromise. We could even spin 1.23 fairly soon
afterwards with scope limited strictly to JDK17 updates.
lewismc
