On 2010/08/02 17:32:43, Ray Ryan wrote:
LGTM
The problem with this approach is that it's so hugely prescriptive
— there is
one and only one way to deal with user auth in RequestFactory. I think
it will
do to get the conversation started in M3, but we'll need to find
something more
flexible before we ship.
Can you tell me more about what you mean here? This CL shouldn't actually have changed anything as far as the ability to configure the User services - the only change is that the base implementation is now checked into GWT rather than generated. In order to use a different authentication method, you would still extend the UserInformation class, and use the web.xml file to indicate that the app should use your class instead. You still have the ability to write your own widgets that use getUserInformationRequest (rather than (or in addition to) using the LoginWidget. Am I missing something else that you had in mind?
http://gwt-code-reviews.appspot.com/721803/diff/1/7 File
bikeshed/src/com/google/gwt/sample/expenses/server/domain/GaeUserInformation.java
(right):
http://gwt-code-reviews.appspot.com/721803/diff/1/7#newcode72
bikeshed/src/com/google/gwt/sample/expenses/server/domain/GaeUserInformation.java:72:
// Does not work - ID is based on the user This should be JavaDoc (and a little bit more descriptive)
Done http://gwt-code-reviews.appspot.com/721803/show -- http://groups.google.com/group/Google-Web-Toolkit-Contributors
