axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Davanum Srinivas <>
Subject Re: Handler changes
Date Wed, 04 Sep 2002 13:27:00 GMT

Checked in the code. your testcase is auto-magically invoked...I dropped the PackageTests
and made sure that the junit test case class ends with the suffix TestCase. Please try out
latest cvs.


--- Greg Truty <> wrote:

> I've made a bit of progress w/the Handler code and have testcase plus some
> code changes (along w/some other bugs that I have noticed).  This is on a
> version of todays driver (9/03).
> The testcase is included below in test\jaxrpchandler.  It tests the
> addition of a client/server side handler, along w/ client-config and
> programatic API access changes.  It deploys/undeploys the service
> appropriately.  The handler adds a header element at each of the hook
> points, incrementing a counter in the header.  Lastly, on the client
> response, it checks to ensure the counter has been updated appropriate and
> calls a static method to the testcase (it's out-of-band) to confirm that
> the handlers executed appropriately.
> The code changes take into account the role/header/parameter options (which
> were wrong/missing in the code orginally checked in).  The CVS diff
> (src\axis.diff) is included in the .zip file provided.
> Lastly, a problem was found w/the support (although it's more the SAAJ
> support) that that the modification of the SOAP body doesn't seem to work.
> It when certain modifications are done using the SAAJ APIs, we occasionally
> mark the child as dirty.  However, in serializing out the whole SOAP body,
> it just checks for the Envelope being dirty.   If I read the code
> correctly, MessageElement has an output method which runs the recorder vs.
> walking through the envelope again (since the SOAPEnvelope is marked as
> dirty).  In addition, modification of body contents (like addition/removal
> of attributes, children, etc doesn't seem to work).  I was trying the case
> where I modified the contents of one of the sub-elements and it doesn't
> know that it's dirty.  I believe that these add/remove attributes/child
> elements also need to set the dirty flag.  This certainly has performance
> implications but I figured that I would start w/providing some basic
> support first.  The specific thing I was trying to do was modify the final
> value in the returned result (which seemed to look good in my debugger...
> but in deserializing the response, it always played back the original
> recorded items).
> After the update, the build, functional tests and all-tests ran w/o a
> problem (however.. .I can't seem to get my testcase to run automagically...
> I will get w/Matt tomorrow and try and figure it out).
> (See attached file:
> Regards... Greg
> Greg Truty

> ATTACHMENT part 2 application/zip

Davanum Srinivas -

Do You Yahoo!?
Yahoo! Finance - Get real-time stock quotes

View raw message