On Wednesday, August 18, 2010 16:41:39 you wrote:
> On Wednesday, August 18, 2010 16:33:04 Ray Cromwell wrote:
> > I have a feeling this is actually working as its supposed to on Chrome,
> > but silently failing on the other browsers. The stack trace indicates
> > that a JSON payload that it thought was a JSO object turned out to be a
> > JS Number, so the JSNI callback failed. Why this would work on other
> > browsers beats me, but I'll track it down in a few hours.
>
> Thing is.. on other browsers, we still get a proper response from the
> RPC request.
>
> I'm currently noticing that r8563 is fixing some of our responses from
> the server. Before this patch, I didn't get any RPC response that would
> actually produce something else than an exception on Chrome. However,
> this is now getting better. I'm trying to see what's the pattern here.
Great.. r8563 did something good. It made the problem very consistent.
Now, I can get it in Firefox. Doing so gaves me a little more details
about what was going on.
It seems that RPC calls returning RequestObject<> are the ones broken.
The 3 interesting frames on the stack when the exception occurs are:
at
com.google.gwt.requestfactory.client.impl.RecordJsoImpl$JsonResults$.getJavascriptResult$(RecordJsoImpl.java)
at
com.google.gwt.requestfactory.client.impl.AbstractRequest.handleResponseText(AbstractRequest.java:100)
at
com.google.gwt.requestfactory.client.impl.RequestFactoryJsonImpl$1.onResponseReceived(RequestFactoryJsonImpl.java:99)
--
http://groups.google.com/group/Google-Web-Toolkit-Contributors
