Is JDK 11+ now supported?

[karllessard]

In this documentation (and maybe elsewhere?), it says that only Java8 is actually supported by MLeap but is it still the case after the upgrade to Scala 3?

I'm asking because in TensorFlow Java, we are planning to abandon the support of Java8 and build our next releases under JDK 11, so I just want to validate if some actions would be required in this project to use the new artifacts, thanks!

[jsleight]

Hey @karllessard thanks for reaching out!

We're only building mleap against java 8 right now so I'm not sure. I'm assuming you meant Spark 3 (not scala 3), and spark 3 does support java 11. MLeap is heavily based on spark but does have additional code which potentially could be incompatible.

So I'm left with an unhelpful answer of "maybe we'd have to make changes but maybe we wouldn't". We do have tensorflow-java as a pinned dependency so if things are incompatible then it just means we'd have to upgrade mleap prior to upgrading tensorflow-java again.

[karllessard]

Yes I meant Spark 3, sorry for that.

We did not started yet to build the JDK11 artifacts but we will soon do. If you want, I suggest that we keep this issue open and I can let you know when we have available snapshots for you to try.

[jsleight]

If you want, I suggest that we keep this issue open and I can let you know when we have available snapshots for you to try.

Sounds good to me 😄

[karllessard]

Hey @jsleight,

The TFJava artifact build on JDK11 are now available as snapshots (See these instructions for how to consume them). We'll continue to maintain the 0.4.x release branch for at least a year to support JDK8 users but only with bug/CVE fixes. The plan is to develop all new feature in next releases on JDK11.

If you want to give it a try, let me know how it goes!

[jsleight]

Happy to announce that I just merged #818 which upgraded mleap to java 11. If folks want to try it out then you should be able to use the snapshot release starting tomorrow