Right now its done very inefficiently, in effect converting between SDO an AXIOM means serializing to a byte array. Have a look at the on going thread "Data flow on a wire", and also http://issues.apache.org/jira/browse/TUSCANY-118
...ant On 3/22/06, Edward Slattery <[EMAIL PROTECTED]> wrote: > > I have done some preliminary work on SDO integration with axis2C, and have > talked to some of the axis2c people about the work. There are two areas I > see that we could be worked on. > > The first is a conversion from an SDO data graph to a tree of AXIOM > objects. > This could be done by taking SDOXMLWriter for inspiration, and replacing > the > actual writing to XML part with a creation of an AXIOM element or > attribute. We could then flow SDOs over axis2c, and provided we had the > same data factory at the other end, could re-build them with an > SDOAxiomReader utility. > I also assume (maybe wrongly) that there is nothing preventing me using > the > SDOAxiomWriter to write XSD information, so we could flow both the data > and > the metadata if required. > > The second area of interest is the guththila parser. I have tested that it > can replace our libxml2 parser, and it seems to be capable of doing that. > We > could investigate having a kind of pluggable parser layer, and switch > parsers between libxml2 and guththila. We would have to work out the best > way of integrating - I.E take a copy of guththila and build with it, or > have > axis2c as a dependency of SDO for C++? The second is cleaner, but I > imagine > the PHP group might want the option of building SDO with only libxml2 for > now. > > I would be interested to know how the integration of the java SDO is being > handled. Anyone aware? > > > On 21/03/06, Pete Robbins <[EMAIL PROTECTED]> wrote: > > > > The current SCA C++ implementation supports webservice binding using > Axis > > 1.x (there are some outstanding problems with this). I would like to > move > > up > > to use Axis2C. > > > > One way to do this is to discard the Axis 1.x binding code and replace > > with > > Axis2C. A better way would be to restructure the code to allow bindings > to > > be more pluggable. As a starting point I think it would be useful, as > > discussed on another thread, to have a Wiki page outlining the current > > code > > architecture/structure which would be of use for anyone who would like > to > > help out. > > > > As part of this it would also be really good to have an SDO C++ > > integration > > with Axis2C. > > > > Pete > > > > > >
