[Webtest] Re: new window

Marc Guillemot webtest@lists.canoo.com
Thu, 13 May 2004 14:21:21 +0200

What would be very nice is to abstract the "WebClient" and configure the implementation to use through properties or 
something like that. This way Webtest would only know the interface. Possibles implementations would be:
- using htmlunit
- using httpunit
- using JRex to drive Mozilla (would be very nice)
- ...

If we go such a way, the difficulty will be to maintain different implementations having different possibilities (cf js 
or windows - not the OS! - support in httpunit). But it would be really great.


Bernhard Wagner wrote:
> Before switching entirely to htmlunit: Could it make sense to let the user
> choose between htmlunit and httpunit?
> I would offer my help to implement the mechanism to make this configurable
> at runtime (I have contributed to a similar infrastructure within httpunit
> to allow for choosing between Tidy and NekoHtml parsers). Sometimes you
> profit from having the choice in particular situations.
> Bernhard
>>-----Original Message-----
>>From: webtest-admin@lists.canoo.com
>>[mailto:webtest-admin@lists.canoo.com]On Behalf Of Dierk Koenig
>>Sent: Donnerstag, 13. Mai 2004 13:56
>>To: webtest@lists.canoo.com
>>Subject: RE: [Webtest] Re: new window
>>Hi Marc,
>>thanks for the verification.
>>For the moment, I'll go for the quick solution and handle
>>_blank specifically. I'll put the clean solution on the list
>>of possible benefits we would have for the move to htmlunit.
>>>-----Original Message-----
>>>From: webtest-admin@lists.canoo.com
>>>[mailto:webtest-admin@lists.canoo.com]On Behalf Of Marc Guillemot
>>>Sent: Donnerstag, 13. Mai 2004 11:49
>>>To: webtest@lists.canoo.com
>>>Cc: Dierk Koenig
>>>Subject: [Webtest] Re: new window
>>>>I'm not absolutely sure, but it seems this behaviour was introduced
>>>>with the JavaScript support for clicklink.
>>>>Marc: could you please verify this?
>>>I guess that you're right.
>>>Before the change for js support, a fully new request was
>>>performed and its result was cosdered as the current page. Now
>>>we let httpunit "click" the link. In the case where a target is
>>>set, it seems that httpunit doesn't update what is
>>>considered as the current page.
>>>>Marc, all: do you have any idea how we can do this cleanly?
>>>The logic would be to consider as current window the one in which
>>>the content of the just clicked link is shown (the one
>>>you see on top when surfing with your prefered browser).
>>>The big problem according to my tests is httpunit behavior: when
>>>calling click() on a link with target="_blank" for
>>>instance, the returned WebResponse is the original one, not the
>>>one called by the link.
>>>Once again: what about htmlunit? With htmlunit this works just
>>>normally and the HtmlPage object has a method
>>WebTest mailing list