Hi Samisa,

I've been delving deeper into this problem for Mark.
We've found that it was not ClientStubWriter.java that introduced the
problem, we have been able to determine it was an update made to
PlatformSpecificUnix.hpp, and was only seen when trace is enabled.
Apologies for wasting your time trying to find a fault in your code.

I have just backed out the change and AxisBench is now passing.
The same fix also resolves problems we're seeing with ExceptionTestDoc and
FaultMappingDoc, which I believe may be the same as the problem you report
in http://issues.apache.org/jira/browse/AXISCPP-555

Regards,
Adrian
_______________________________________
Adrian Dick ([EMAIL PROTECTED])



                                                                           
             Samisa Abeysinghe                                             
             <[EMAIL PROTECTED]                                             
             usa.com>                                                   To 
                                       Apache AXIS C Developers List       
             21/03/2005 09:49          <[email protected]>          
                                                                        cc 
                                                                           
             Please respond to                                     Subject 
              "Apache AXIS C           Re: Samisa - please help me         
             Developers List"                                              
                                                                           
                                                                           
                                                                           
                                                                           
                                                                           




Hi Mark,
             Fortunately I had a backup of the file prior to making
changes. PFA.
             One of the problems of identifying the changes is because I
did some
indentation fixes. Sorry about that.
             I ran the test on Linux and Chinthana helped me to test this
on
Windows. On both platforms, with the changes in place, AxisBench client
does not crash. However, the output we get is different from the
formatting in the expected out file (but the outputs seem correct)
             When you say fail does it mean that it segfaults? Are there
any other
tests that segfault?
Thanks,
Samisa...

On Fri, 2005-03-18 at 17:27, Mark Whitlock wrote:
>
>
> Hi Samisa,
> I am investigating AxisBench failing on linux. It appears to fail when
> deleting the last (i=99) output BenchBasicDataType. This test succeeded
on
> the 14th March but failed on the 15th (and all builds subsequently). So I
> looked through all the changes that were made on the 14th and 15th March
> and I have eliminated all but your change to
> cpp/literal/ClientStubWriter.java. Unfortunately the file got formatted
for
> this change so it is hard to see exactly which lines you really changed.
In
> your JIRA comment you said you had to make large changes.
>
> I apologise if I am unfairly accusing your fix. I just need some help
> sorting this problem. I will try (privately) backing off your change and
> see if that fixes AxisBench but that may be difficult since code has been
> added since.
>
> Please could you help?
> Thanks,
> Mark
> Mark Whitlock
> IBM
--
Samisa Abeysinghe <[EMAIL PROTECTED]>
Virtusa Corporation
(See attached file: ClientStubWriter.java.0.ind)

Attachment: ClientStubWriter.java.0.ind
Description: Binary data

Reply via email to