Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [glassfish-dev] Will GlassFish 7.0 support running on Java SE 17?


On 5/4/22 6:01 PM, Scott Marlow wrote:


On 5/4/22 3:59 PM, arjan tijms wrote:
Hi,

On Wed, May 4, 2022 at 9:42 PM Scott Marlow <smarlow@xxxxxxxxxx> wrote:

Will GlassFish 7 support running on Java SE 17? 

Definitely, much of the development from the last few months have been focussed on running on JDK 17 and recently 18.

We have quite an amount of embedded tests covering basically all APIs, and they all pass on JDK 18 (and 17 and 11): https://ci.eclipse.org/glassfish/view/GlassFish/job/glassfish7_build-and-test_jdk18

All the standalone TCKs that I tried passed on JDK 17, and some on JDK 18 as well.

I wonder why specifically the platform TCK fails so many tests. Isn't it perhaps just the add-opens in many cases? If that isn't set, many tests indeed fail, so this single option is quite important.

https://ci.eclipse.org/jakartaee-tck/job/jakartaee-tck/job/master/1774 had 9058 test failures running Java 17, likely too many to examine without automation.  To make it worse, there are much noise like:

"
javax.net.ssl.SSLException: java.io.FileNotFoundException: /.gfclient/truststore (No such file or directory)

"

However the good news is that with ^ test run, I don't see the the [2] failure from an initial look.  Still, we have 9058 failures.  Thanks for the answer about Java SE 17, so we can look more closely at the output of https://ci.eclipse.org/jakartaee-tck/job/jakartaee-tck/job/master/1774.  The job didn't collect artifacts from each separate job so we just get the big 1 gig file.  I will start a new test run with fewer tests to see if that is better.

I started a EE 10 smaller test (only running `ejb30/lite/singleton`) run with Java SE 17 and the first test failure is caused by a deployment failure.  The deployment failure is preceded by a `java.lang.reflect.InaccessibleObjectException:Unable to make field private static javax.naming.spi.InitialContextFactoryBuilder javax.naming.spi.NamingManager.initctx_factory_builder accessible: module java.naming does not "opens javax.naming.spi" ` + `IllegalStateException: InitialContextFactoryBuilder already set`.

https://gist.github.com/scottmarlow/9d51e0c7e306c02070ed5e6c8cba0453 shows the `InaccessibleObjectException` error first #1 as that is the first error that occurs, the deployment failure is shown as #2.

Scott


Scott


_______________________________________________
glassfish-dev mailing list
glassfish-dev@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/glassfish-dev

Back to the top