Orekit unit test fails

Orekit development
#1

Dear all,

We recently found a small, but annoying, bug. This bug only impacts the Orekit tests. In other words, it does not impact the main library.
Details about the bug are given here: https://gitlab.orekit.org/orekit/orekit/-/issues/744

To summarize: Since the 1st January 2021 an Orekit’s unit test fails. It fails because the value we want to verify in the test changes each year. Therefore, the value changed the 1st January 2021. The value is good but it does not correspond to the reference value calculated for 2020.

Fixing this bug seems to be easy. However, I have to ask you two questions.

Is the bug impacting your applications using Orekit?
Do you think we must create a patch? Or the fix can wait release 11.0, since it just impacts the tests.

Best regards,
Bryan

#2

As the library itself is fine and the failure is limited to an ill-conceived test,
I think it can wait for next version.

It should be fixed in the develop branch, though.

Let’s wait for other opinions about that, I will ping the PMC to see if they think a patch release 10.3.1 is needed. Anyway, for patch like this, the governance allows for a fast track release without PMC vote formally required, so we could do it if we want.

#3

I completely agree with Luc.
The bug only concerns a badly designed test, it should be fixed in the develop branch, but it does not affect the library itself, so I don’t think it requires a specific patch release.

#4

Hi everyone,

No impact on our side. I do not think that this minor issue requires a hotfix, waiting for the next release should be fine.

#5

I agree with all of you. Let’s fix it for next version since it is not in the library itself but in the tests.

#6

Great, I also agree with all your remarks. Since it is a bug in the tests, the fix can be done for the next release.

#7

May I ask what functionality was involved in this test?

Paul Cefola

#8

Dear Paul,

The falling test is related to CR3BP model. It is a model for three-body problem. For instance, to model trajectories around Lagragian points.

Bryan

#9

Issue is now fixed into develop branch

Bryan