Hi Dennis; Actually real interst of Axis2 was to use a *Typed* Stax implementation! But maybe now we have deviated too much by now.
We can convert the typed values to strings and use them with our current model. But then we do not gain by using a typed represnetation. Others what do you think? Thanks Srinath On 7/14/05, Dennis Sosnoski <[EMAIL PROTECTED]> wrote: > Hi Srinath, > > I understand the approach, but just don't see how you can do that with > DFDL. DFDL describes structured binary data. You could probably do some > sort of SOAP-like protocol variation to support it, but AFAIKS it > wouldn't be using anything like StAX - it'd require a typed binary > interface, rather than an XML Infoset interface. DFDL seems much closer > to Sun's "FAST" Web Services project, which uses ASN.1 encoding with > binary transfers of object data (as opposed to the Fast Infoset version, > which is an encoding of XML data). That seems to require much deeper > structural changes than just using a different StAX reader/writer > implementation. > > - Dennis > > Srinath Perera wrote: > > >Hi Dennis; > >Just to add some info, we are thinking about adding the binary supportby > >writing a impl forStax readers and writers works on top of binary protocol. > >As Axis2reads and writes using stax .. will not notice any differance. > >ThanksSrinath > >On 7/13/05, Dennis Sosnoski <[EMAIL PROTECTED]> wrote:> Hi Mark,> > I've > >been ignoring the Axis/Axis2 lists for a while, so I may have> missed some > >earlier discussions on this - can you give the reason for> choosing these > >two approaches in particular? I'd personally love to see> a comparison > >between the XBIS format I originated (http://www.xbis.org)> and Fast > >Infoset. Both of these deal with XML-based representations,> while DFDL (as > >I understand it) is strictly binary. Since DFDL doesn't> really relate to > >XML at all it seems like it'd be difficult to fit it> into the Axis2 > >structure (including AXIOM) in any reasonable manner.> > - Dennis> > Mark > >Pimentel wrote:> > >Hi everyone,> >I'm Mark Pimentel, a computer engineering > >student from CarnegieMellon, and I'm currently doing a Google Summer of > >Code(code.google.com) project with Apache Axis 2 on binary XMLserialization. > > My original proposal is located online > >athttp://www.andrew.cmu.edu/user/mpimente/MarkPimentel-SummerOfCode.pdf. My > >project mentor for this project is Eran Chinthaka.> >I had some delay in > >getting started, so I've had to push some itemsback on my weekly schedule. > >The updated schedule is as follows:> >> >WEEKLY SCHEDULEMark Pimentel, > >Google Summer of Code 2005> >Week of July 10thConclude research of different > >formats. Will submit a documentoutlining the major aspects of the two > >representations that I will belooking into—Fast Infoset & DFDL. Am > >currently targeting Fast Infosetas the first schema to implement, and DFDL > >as the second.> >Week of July 17th Begin implementing a StAX parser to write > >binary XML data using theFast Infoset standard. This week, I will > >concentrate on writing theparser and achieving some type of module with > >binary serializationfunctionality.> >Week of July 24thWill continue to adapt > >parser to handle the Fast Infoset standard. Most of the work this week will > >be towards expanding supportedlanguage.> >Week of July 31thThe goal for this > >week is to clean up the first parser code, andperform more extensive > >testing. I will also prepare any progressreport information needed by the > >Google Summer of Code program.> >Week of August 3rdBegin development of > >parser for DFDL standard. If similar parts ofthe parser from the first can > >be used, then work from later weeks canbe started.> >Week of August 14thWill > >continue fleshing out and expanding DFDL support.> >Week of August > >21stAiming for more definite testing and verification of second parser. If > >time permits, I will also start on a third parser.> >Week of August > >28thImplementation of third representation, if possible. Otherwise,return > >to first two parsers to ensure correctness and correct anyremaining > >integration issues.> >> >Thanks,Mark> >> >> >> > > > > > > > >
