axis-c-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Samisa Abeysinghe <>
Subject Re: [AXIS2] OM serialization/deserialization working
Date Tue, 25 Oct 2005 08:34:38 GMT
Paul Fremantle wrote:

> Samisa
> Congratulations. That is impressive!
> Paul
> On 10/21/05, *Samisa Abeysinghe* < 
> <>> wrote:
>     Hi All,
>         After much effort, I am happy to tell you that we have a running
>     version of OM in SVN.
>         We can give an XML stream and let the builder build the OM
>     structure. Also, we could build the OM tree manually. Once built, OM
>     structure can be serialized. In short most of the operations
>     mentioned
>     in could be done in C.
>         Initial tests are successful. There are some memory leaks.
>     Also full
>     functionality is yet to be verified.
>         Environment (with allocator, string, log, error and stream)
>     has been
>     used in this implementation.
>     TODOs:
>     1. Complete the doxygen comments - I am working on this currently and
>     hope to complete soon. For all the environment related stuff
>     (allocator,
>     error, etc.) doc comment are complete.
Doc comments are now in place and doxygen can be used to generate the 
docs for the Util module and the OM module source headers.

>     2. Need to handle logging. The code has provision to handle logging.
>     Error handling has been done, but related logging code is yet to
>     be written.
This is a TODO, but as the error handling is in place, this can be done 
any time one feels doing so.

>     3. Fix the memory leaks
There are some *nasty* and some *very bad* leaks in the code. I used 
valgrind and the picture is somewhat grim. I managed to fix some, but 
some more critical leaks exist. Bulk of the leaks are related to 
guththila implementation. I think it is worth spending some time right 
now to fix these leaks.
The best is to check leaks on guththila first and then look into leaks 
in axis2 code with guththila.

As the OM code is almost complete, it is right time for code to be 
reviewed for bugs and suggestions. Early comments would be of great help.


>     4. Verify the possibilities for differed building. The
>     understanding is
>     that this would be possible with current design but that has to be
>     verified.
>     I would like to thank Nandika for his contributions to OM code as
>     well
>     as to Denesh for his initial Guththila implementation and help with
>     troubleshooting and bug fixing.
>     Thanks,
>     Samisa...

View raw message