Numerical differences in test

Hi folks,

I recently discovered that some tests fail with java 8, but not on the pipeline which uses 11. I opened an issue on one in particular before realising the scope was wider.
There seems to be some numerical differences. I don’t know if this just appeared with @Ryan’s contribution (delegating more to Math) or if it is older.

Anyway, that’s just another symptom of still using old Java. Could we possibly trigger the Hipparchus PMC vote for the next release to be in Java 11?
In the wake of this, the Orekit PMC could then vote to have Orekit 15.0 as a “twin” of Orekit 14.0 but in Java 21, as discussed in the relevant threads.

Cheers,
Romain.

2 Likes

Yes, I think we could do that.
We could even decide 13.X in Orekit will be the last using Java 8 and upgrade to Java 21 directly for Orekit 14.

I have also seen recently that recent versions of Android are compatible with Java 17, perhaps that could be suitable for Axel needs on Hipparchus?

1 Like

We could even decide 13.X in Orekit will be the last using Java 8 and upgrade to Java 21 directly for Orekit 14.

Sounds good to me.

@axelclk is java 17 more acceptable now?