On Thu, 2 Aug 2001, Stephane Bailliez wrote: > mmm... ? what will be included from common if the cleanup is not performed > in all releases ? I mean if crimson already has subset(DOM2) + SAX2 and that > Xalan has JAXP.bad + subset(DOM2) + SAX2. > Do you plan to perform the lean distrib for crimson and xalan yourself ?
jaxp.good and xml.common are virtually identical. jaxp from xalan is very close ( it uses introspection instead of a proxy to call context class loader, it's just a different trick ). Any of those should work. And of course, I hope we'll release xalan with the jaxp from commons - everyone agrees, we just need to do it. > What is especially problematic to me is JAXP. Assuming you have > JAXP.good.jar and JAXP.bad inside xalan.jar > What is your guarantee that JAXP.good will be before xalan.jar when loading > them unless you specify the order ? Yes, and think of the fun when you use JDK1.4 :-) ( well, the good part is that a jaxp.good will be included in the system class loader ). Well, I understand your point, and I'll try to fix this where I can. > will not be used. But I guess this is normal since they did not have time to > rethink about their classloader hierarchy like you did for Catalina. :) Please ! I have nothing to do with Catalina ( and I'm not sure about the "think" part in their classloader scheme ). I did part of the 3.3 classloader hierarchy, and I'm quite happy with that. Costin
