LGTM On Tue, May 31, 2011 at 1:39 PM, <jbrosenb...@google.com> wrote:
> > > http://gwt-code-reviews.appspot.com/1450806/diff/3002/dev/core/src/com/google/gwt/core/ext/GeneratorContext.java > File dev/core/src/com/google/gwt/core/ext/GeneratorContext.java (right): > > > http://gwt-code-reviews.appspot.com/1450806/diff/3002/dev/core/src/com/google/gwt/core/ext/GeneratorContext.java#newcode34 > dev/core/src/com/google/gwt/core/ext/GeneratorContext.java:34: * likely > to succeed. > On 2011/05/31 16:27:01, rjrjr wrote: > >> Why is the argument a string rather than, say, JClassType? >> > > > Since this method will be called from within a generator, there may not > be a valid JClassType from type oracle (but we can check for the > availability of a rebind rule, by sourceTypeName, which is how they are > expressed in rebind rules). The actual type might come into existence > after the current (and other) generators complete. > > > Wouldn't the second sentence would be more accurate as: "is likely to >> > succeed > >> for a type with no default constructor. (Any concrete type with a zero >> > args > >> constructor can be instantiated via GWT.create().)" >> > > > Can't really check whether a concrete type with a zero args constructor > might be available, since it might only come into being after other > generators run. > > > Also, it isn't GWT.create(sourceTypeName), the argument is a class >> > literal. > > Good point. I've removed references to GWT.create() altogether in the > javadoc. > > > http://gwt-code-reviews.appspot.com/1450806/ > -- http://groups.google.com/group/Google-Web-Toolkit-Contributors