2008/8/13 Kai Mütz <[EMAIL PROTECTED]>
> [EMAIL PROTECTED] <> wrote:
> > Hi,
> >
> > [back from vacation]
> >
> > I'm sorry that some regression has been introduced in HtmlUnit-2.2
> > despite the high test coverage of HtmlUnit itself as well of different
> > projects using HtmlUnit (WebTest of course, but JSFUnit, JWebUnit and
> > WebDriver).
> >
> > Can you provide a (simple) example allowing to identify the
> > regression introduced with HtmlUnit-2.2?
>
> Currently I can not provide a simple example but can try tu explain our
> problem:
>
> We have a webapplication running in an frameset with a header, a left and a
> main frame. When a page is loaded in the main frame onload event executes a
> JS function located in the left frame in order to update the content of the
> left frame. The JS function is something like:
>
> function loaded(document) {
> if (0 == parent.frames.length) {
> return;
> }
> [....]
> if (parent.left) {
> [....]
> parent.left.loadHistory();
> }
> }
>
> If webtest invokes one of these pages (with <invoke/> step) it is loaded
> into the entire window not into the main frame (correct me if I am wrong).
> Thus no frameset exists, thus no "left" frame exists. In this case webtest
> R_1714 throws an error:
>
> TypeError: Cannot find function loadHistory in object [object]
>
> This is strange bacause the JS checks if a left frame exists. This error
> does not occur in R_1713, Firefox 2 and 3, IE 6 and 7.
>
> I think it is a HtmlUnit-2.2 issue.
>
>
Considering other webtest errors with R_1714 yields me to the conclusion
that it must have something to do with the frame handling of HtmlUnit-2.2.
HTH, Kai
