Code Review Request: (JAVASERVERFACES-3593) Navigation from method-call-node to method-call-node does not work

From: zhijun Ren <>
Date: Thu, 15 Jan 2015 14:22:46 +0800

Hi Ed,

I have attached change bundle to the jira issue. The change is big and
including 25 cases to cover all the navigation possibilities. I have
discarded your original tests because of the complete coverage.

I noticed that the flow can be defined via xml file and java annotation,
so I don't know whether need add more cases to test the combinations of
the navigation test with different flow definition ways.

Please help to review due to the big code size and I am looking forward
your comments.

I have tested all the 25 cases against GF and get the following results
and I think the failed tests may be caused by the Mojarra implementation
code and I am checking into it.

Time elapsed: 0.045 sec <<< ERROR!
com.gargoylesoftware.htmlunit.FailingHttpStatusCodeException: 500
Internal Server Error for

Tests run: 5, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 0.165
sec <<< FAILURE! - in
Time elapsed: 0.032 sec <<< FAILURE!
java.lang.AssertionError: null
     at org.junit.Assert.assertTrue(
     at org.junit.Assert.assertTrue(

Results :

Failed tests:
   SwitchNaviToOtherNodesIT.testNaviToReturn:112 null

Tests in error:
FailingHttpStatusCode ...

Tests run: 25, Failures: 1, Errors: 1, Skipped: 0/


On 1/14/15, 1:43, Edward Burns wrote:
>>>>>> On Wed, 7 Jan 2015 02:24:54 +0000 (UTC), " (JIRA)"<> said:
> ZR> I have ported your changes into my local trunk workspace and run all
> ZR> the tests for glassfish.
> ZR> The newly added uncompleted tests failed. To complete them, I need
> ZR> to get familiar with the jsf flow.
> There are some failures, yes. This is because the bug is basically,
> "make all possible flow traversals work". My process for fixing the bug
> is using test driven development. As I worked through all the possible
> flow traversal combinations I discovered more problems.
> ZR> So I need more time to do it, please stay tuned.
> Yes, that's understandable. It might be better to do the work in stages
> and commit the parts that currently do work, then file one or more
> separate bugs for the remainder.
> Ed