Bug 3566

regression test is pending for after x.6.8

test is in svn

Added Iliyan. Regression test is on svn head and fails now there

Vladimir, is this one related to your recent svc.conf changes?

(In reply to comment #4)
> Vladimir, is this one related to your recent svc.conf changes?
> 

Yes, this one is related but the test itself is a bit wrong. I'll check it.

(In reply to comment #5)
> (In reply to comment #4)
> > Vladimir, is this one related to your recent svc.conf changes?
> > 
> 
> Yes, this one is related but the test itself is a bit wrong. I'll check it.
> 

Thanks, 3565 is also related, that test in the repository needs an update also

(In reply to comment #6)
> Thanks, 3565 is also related, that test in the repository needs an update also
> 

I don't see a difference between tests for 3565 and 3566. I think the following code in 3566 is wrong and needs to be fixed.

second_extra[2] = ACE::strnew (ACE_TEXT ("-ORBSvcConfDirective \"static Client_Strategy_Factory '-ORBReplyDispatcherTableSize 666'\""));

Each command-line option has to be a separate element in argv array. So, -ORBSvcConfDirective and \"static Client_Strategy_Factory '-ORBReplyDispatcherTableSize 666'\" has to be 2 elements. Once this is fixed 3565 and 3566 will become equal.

it is a long time we made this test, for 3565, the test in the repository needs an update when it gets fixed, see the bugzilla issue. the quoting in this test should be ok so far as I remember, without -ORBGestalt local, the quoting is not a problem

(In reply to comment #8)
> it is a long time we made this test, for 3565, the test in the repository needs
> an update when it gets fixed, see the bugzilla issue. the quoting in this test
> should be ok so far as I remember, without -ORBGestalt local, the quoting is
> not a problem
> 

Without -ORBGestalt local I see the same error as with it.

(18552|140737353955088) EXCEPTION, Exception caught:
system exception, ID 'IDL:omg.org/CORBA/BAD_PARAM:1.0'
TAO exception, minor code = 11 (ORB Core initialization failed; EINVAL), completed = NO

I can't remember the details, but on the commandline I thought you could quote the directive in such a way that it is just one argument. Not sure about the details, as far as I remember the quoting was not maintained in some places. Feel free to update the test with 2 arrays and than mark this and 3565 as fixed, 

I think the test itself is fine, I will build 1.6.8 on a system to see what I got there