Jim, I can see your point. On the other hand, unless xjc can be coopted, wsdl->aegis will be a really big project. The doc I've found on xjc is really uninformative, so I have not been able to convince myself that it could be persuaded to write additional files. I suppose that if it would write additional comments we could post-process.
I have a strategy in mind if we decide to go here. Step 1 is to define Aegis as being complementary to JAXB instead of completely parallel. Then, step 2 is to implement ->aegis by concentrating on the cases you cite with real value, instead of a complete recreation of xjc. --benson > -----Original Message----- > From: Jim Ma [mailto:[EMAIL PROTECTED] > Sent: Sunday, September 23, 2007 11:50 PM > To: [email protected] > Subject: Re: Is the idea of wsdl2java (aegis) really useful? > > Hi Benson , > > As far as I can see , aegis data binding is complementary to JAXB > whether from java-> wsdl or > wsdl-> java direction . > > I have not dug into aegis data binding code to find out if it can do the > following things ,but i think > having these feature in wsdl2java makes sense : > 1. Generating simple code-first-liked java class that easy to understand. > 2. Generating java.util.Data type without add a XmlTypAdaptor > customization to map xsd:dateto java.util.Date > and other thing we need to use jaxb customization to map a shcema > type to a common jdk class. > And also we can use aegis data binding in wsdl first direction as it's a > faster and StAX based data binding . > > Since there are some data types that aegis does not support , maybe > we can add this support in wsdl2java > later. > > Regards > > Jim > > > > > Benson Margulies wrote: > > I trust that our colleagues on the far side of the Pacific are reading > > this thread and will chime in if they disagree, as they were the ones > > who asked me if I intended to proceed to the 'other direction,' I think. > > > > > > > > ________________________________ > > > > From: Dan Diephouse [mailto:[EMAIL PROTECTED] > > Sent: Saturday, September 22, 2007 4:14 PM > > To: [email protected] > > Subject: Re: Is the idea of wsdl2java (aegis) really useful? > > > > > > > > Yeah, I think its definitely a waste of time to do an Aegis > > schema->java. We would gain absolutely no benefit from it as far as I > > can tell :-) > > > > - Dan > > > > Daniel Kulp wrote: > > > > I had this conversation with Dan D. before and the thoughts were pretty > > much exactly the same. If you are generating java code from schema, > > you should use the "standard" JAXB stuff since that supports more of the > > > > schema types than Aegis does. Aegis is primarily targeted for the > > Java first case. Having java -> wsdl tools for it makes sense. > > However, the wsdl -> java stuff makes less sense. > > > > Dan > > > > > > On Friday 21 September 2007, Benson Margulies wrote: > > > > > > I'm wondering why anyone would want this. If the user is going > > to run > > a code generator, what's wrong with JAXB? The virtue of Aegis is > > to > > avoid an infestation of snails (@) in existing code. Once you > > are > > firing up a generator, it's new code. > > > > > > > > I note that Aegis has some not-very-documented capabilities, and > > I > > wonder if some of them are superior to JAXB in some way. > > > > > > > > Do we ever post some sort of [POLL] to the user list asking 'who > > would > > use this feature' questions? > > > > > > > > > > > > > > > > > > > > > > > > > >
