Am 2020-01-02 um 19:47 schrieb Oleg Kalnichevski:
On Wed, 2020-01-01 at 23:46 +0100, Michael Osipov wrote:
Am 2020-01-01 um 14:45 schrieb Oleg Kalnichevski:
Folks
Shall I proceed with HttpCore 5.0-beta11 release or not?
Do we still need to rename anything in core?
One nit during testing which should be addressed before GA:
When running: JSSEProviderIntegrationTest and Conscrypt JAR does not
contain a JNI wrapper for the platform the tests are run:
unzip -t ./org/conscrypt/conscrypt-openjdk-uber/2.2.1/conscrypt-
openjdk-uber-2.2.1.jar | grep native
testing: META-INF/native/ OK
testing: META-INF/native/libconscrypt_openjdk_jni-linux-
x86_64.so OK
testing: META-INF/native/conscrypt_openjdk_jni-windows-
x86.dll OK
testing: META-INF/native/libconscrypt_openjdk_jni-osx-
x86_64.dylib OK
testing: META-INF/native/conscrypt_openjdk_jni-windows-
x86_64.dll OK
tests fail with:
> JSSEProviderIntegrationTest.*[Conscrypt TLS v1.2] »
UnsatisfiedLink
> JSSEProviderIntegrationTest.*[Conscrypt TLS v1.3] »
UnsatisfiedLink
in detail (example):
[ERROR] testSimpleGet[Conscrypt
TLSv1.2](org.apache.hc.core5.testing.nio.JSSEPro
viderIntegrationTest) Time elapsed: 0.34 s <<<
ERROR!
java.lang.UnsatisfiedLinkError: no conscrypt_openjdk_jni-freebsd-
x86 in java.lib rary.path
That's bad portability. I would expect an Assume which verifies that
the
security provide has been successfully loaded.
Any recommendation how that could be accomplished?
No, not yet. I will file an issue first to track this and not to be
forgotten. There is a workaround to disable this test for now.
Michael
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
