Dan, Thanks. I have a patch for 1043 that tries the defer trick in this particular case of maps, without trying it as the general strategy. If I can prove that it's a monotonic improvement on the situation I'll post it to the JIRA.
I'm not in any special hurry to have this stuff. I just keep pulling threads in the Aegis sweater and finding myself with a herd of sheep. It's kind of hard to decide when to stop -- every time I try to make a test case to regress one little thing, I find some other problem. ==benson > -----Original Message----- > From: Dan Diephouse [mailto:[EMAIL PROTECTED] > Sent: Thursday, September 20, 2007 8:45 AM > To: [email protected] > Subject: Re: CXF-1043: should Aegis blend java5 and XML? > > Hi Benson, > > I think that would work. When I originally wrote Aegis it was the other > way around - Java 5 delegates to XMLTypeCreator which delegates to > DefaultTypeCreator. This was because XFire was Java 1.4 and we had Java > 5 as another module. But I could see how this would be useful :-) > > I'm having troubles with my internet right now so for some reason I > can't go through the JIRA issues, but I'll be looking at all of your > issues early next week and will try to get as may fixed as possible (or > apply your patches if you've supplied them)! > > - Dan > > Benson Margulies wrote: > > CXF-1043 complains that, as soon as a property or parameter has some > > XML, that everything has to come from the XML. What I expected was that > > Java5 information would be used to fill in whatever the XML wasn't > > talking about - particularly the generic code. > > > > > > > > There's an obvious opportunity to make the XMLTypeCreator start with the > > info from the 'nextCreator' and then override. Does anyone else have any > > opinions on this idea? > > > > > > > > > -- > Dan Diephouse > MuleSource > http://mulesource.com | http://netzooid.com/blog
