Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [jersey-dev] Jersey build vs. JDK version


just a small update on the JavaFX dependency. I just checked this in detail. Actually it is just a single JavaFX class (javafx.util.Pair) which is used in a test. I guess this should be easy to remove. I'll create a pull request later today.


Am Sa., 5. Mai 2018 um 13:07 Uhr schrieb Markus KARG <markus@xxxxxxxxxxxxxxx>:

I'm also willing to help fixing that, but we need a clear statement of the project lead and the committers what the solution shall look like. For example, I'm +1 for getting rid of any dependency on JavaFX and OracleJDK, as JAX-RS has nothing to do with neither Oracle nor GUIs. But according to the EF rules, it is up to the committers to decide about that. So I kindly like to ask the committers to allow us to simply remove such dependencies.



From: jersey-dev-bounces@xxxxxxxxxxx [mailto:jersey-dev-bounces@xxxxxxxxxxx] On Behalf Of Christian Kaltepoth
Sent: Samstag, 5. Mai 2018 10:32
To: jersey-dev@xxxxxxxxxxx
Subject: [jersey-dev] Jersey build vs. JDK version


Hi all,


it looks like there are many people (including me) who run into problems building Jersey.


The problems I'm aware of are:

  • Jersey uses JMockit 1.18 which has a known bug that causes build failures with OpenJDK. I already created a pull request that updates JMockit to 1.19 which fixes this problem.
  • The "e2e-entity" module uses JavaFX classes. As OpenJDK doesn't include JavaFX, the build fails on this module. Perhaps this module could only be included when running with an Oracle JDK? This should be easy using a Maven profile which gets activated depending on the JDK version.
  • The LinkingTest from the "declarative-linking" module fails when running the build with the latest OpenJDK 8 version. It looks like the order of the links returned from the resource is different and the test is (too?) strict about checking the response. Disabling JSONAssert's strict mode fixes this problem. I'll create a pull request for this later today.
  • Markus reported that ClientDestroyTest fails for him when running the build with Oracle JDK 1.8.0_144
  • On the mailing list Jan told us that there are known problems with JDK 8b161+ 

So you see that running the Jersey build is not very straight forward. :-)


So I wonder if this could be improved. I think that the issues regarding newer Oracle JDK versions are definitely something that should be fixed. What about OpenJDK support? Is there any specific reason for relying on Oracle JDK at the moment?


I would love to help the Jersey team to improve the situation! So please let me know how to help.





jersey-dev mailing list
To change your delivery options, retrieve your password, or unsubscribe from this list, visit


Back to the top