hstaudacher / osgi-jax-rs-connector

An OSGi - JAX-RS 2.0 Connector, software repository available on the link below
http://hstaudacher.github.io/osgi-jax-rs-connector
Other
190 stars 98 forks source link

Test 4.2.0 RC2 #68

Closed hstaudacher closed 9 years ago

hstaudacher commented 9 years ago

The RC2 is ready to be tested. There was a critical bug in RC1. If no one complains I will do a release on Friday.

Maven: https://oss.sonatype.org/content/repositories/snapshots/ ziped p2 repo: https://dl.dropboxusercontent.com/u/5808972/com.eclipsesource.jaxrs.repository-4.2.0-RC2.zip

@kaikreuzer can you test it again with openHAB?

kaikreuzer commented 9 years ago

I've tested RC2 and can confirm that the NPE has indeed disappeared and all works fine - in my IDE setup. Unfortunately, testing RC2 on my RaspPi still shows exactly the same problem as described here: https://github.com/hstaudacher/osgi-jax-rs-connector/issues/67#issuecomment-65857745

I will have to investigate, how this can be reproduced on a normal PC. Will try to do so the next days - but better don't plan the 4.2 release for Friday!

hstaudacher commented 9 years ago

That's so strange. It works for our applications as expected. I can't reproduce the error.

In the meantime I will take another look at #69.

kaikreuzer commented 9 years ago

Hm, I do not really get any further :-( I have tried remote debugging and I have seen that the Jersey does not know about the GsonProvider inside the WriterInterceptorExecutor, so somehow this registration seems to fail. I tried to tweak around with the start levels of the different bundles, but could not get it to work on the RaspPi. It is important to note that exactly the same binary works flawlessly on my Mac. So it must be some concurrency issue that only happens, if the start is very slow (90s on RaspPi vs. 5s on my Mac). Is there a given start order that needs to be respected? Should all bundles be auto-started or better not? Please give me some advice what I could check further, I do not really have many more ideas...

hstaudacher commented 9 years ago

The order should not matter because the connector gets all service registrations. I have created a separate issue for this, see #72

I try to produce a standalone example that I can run on a pi. But I will have access to a pi earliest next week wednesday.

But after I have created the example maybe you can validate that the error still exists on your pi using it? I will make it available to download.

hstaudacher commented 9 years ago

Will close this issue because RC2 can't be used because of bug #72