[
https://issues.apache.org/jira/browse/AXISCPP-654?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
nadir amra resolved AXISCPP-654.
--------------------------------
Resolution: Fixed
Fix Version/s: current (nightly)
Well, it may have been a minor improvement but it resulted in major changes:
-- removed duplication in xerces parser and this all the heavy duty logic is in
one routine that all the others call.
-- the xerces parser keeps all the state information, and throws exceptions on
SAX or XML or transport exceptions. In addition, it keeps a copy of the error
information for subsequent retrieval by new getErrorString() and
getErrorCode() methods. At this time there code to keep track of peeks and
whether to return NULL string on start/end elements. So it makes the code a
little more complex that I would like. I think that logic can be removed and
peek() can be eliminated by updating SoapDeSerializer.cpp, but I first have to
figure out when a node is consumed or not - i.e. some literal logic in there
uses m_pNode if it is set, while for encoded it does not even look at the
m_pNode, it just gets the next node available from the parser. I think that
can be cleared up, but first I wanted to put the code as-is into SVN.
-- Once I got done with parser, I had to overhaul SoapDeserializer, so while I
was in there I removed a lot of duplicate code and removed unnecessary methods.
For example, I was able to eliminate following methods:
getCmplxFaultObjectName()
decodeFromBase64Binary()
decodeFromHexBinary()
deserializeLiteralArray()
deserializeEncodedArray()
deserializeLiteralComplexArray()
deserializeEncodedComplexArray()
and still keep the methods managable enough, and in some cases smaller than it
was previously. Because the parser now throws exceptions, skipNode() and
getNextNode() are new methods that are called to get parser data and all the
code now calls these routines except for anyNext() (one method uses this).
skipNode() call getNextNode(). getNextNode() can be called with the option of
suppressing the throw'ing of an exception. I also had to put try/catch blocks
in some areas of the code in order to free up any allocated memory before
re-throwing the exception. Now if I can figure when I should mark m_pNode as
consumed, I can probably get rid of peek in the parser and make the parser code
more simpler.
So go at it.
I am marking this as resolved until I get some testcases out there that test
mal-formed SOAP messages.
> Improve XercesHandler failure notification
> ------------------------------------------
>
> Key: AXISCPP-654
> URL: https://issues.apache.org/jira/browse/AXISCPP-654
> Project: Axis-C++
> Issue Type: Improvement
> Components: Client - Deserialization, Server - Deserialization
> Reporter: John Hawkins
> Priority: Minor
> Fix For: current (nightly)
>
>
> The XercesHandler gets called whenever there is a failure or warning -this
> would be an ideal place to put some trace or FFDC !
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
