Marc,

OK .. we've come up with a workaround for this problem.  What we did was
just put a null function called CollectGarbage() at the top
of every page.  So now it's no longer undefined.  One the engineers here
told me it was a little trick MSFT pulled for some ActiveX
Excel stuff.  It's useless, and now merely serves to break things.  Still
this should not trip up WebTest the way it did.  We should
be able to say something like "OK, undefined.  Ignore it and move on."

Unfortunately, I now have another problem that is, probably, a symptom of my
naivete.  I'm running invoke on an url like this:

http://www.bebo.com/InviteJoin.jsp?Member=N&ABC=Y

This causes the test to die immediately, with a message to the effect that
'ABC' must be followed by a ';' delimiter.  So, I put the
semi-colon delimiter at the end, and it makes no difference.  I get the same
error.  The I remove the '=' and just put ABC;.
Now the error goes away but of course I don't get the parameter that I want
set.  Should I be using a POST method here with
a content string?  This is not at all clear to me from the documentation.

Sorry for the elementary nature of this question.  Maybe I'm just getting
tired or something.  Probably, someone else on this list
will know the answer immediately??



On 3/4/08, Ken Hopf <[EMAIL PROTECTED]> wrote:
>
> Hi all,
>
> John & Pip .. thx!  I too hope I can get this thing to work.
>
> Actually, I should have included the full details of the error report for
> Marc.  Don't know what I'm thinking.
> Here it is:
>
> Details javascript errorReferenceError: "CollectGarbage" is not defined. (
> http://s.bebo.com/script/mootools.v1.11.js#3(eval)#1<http://s.bebo.com/script/mootools.v1.11.js#3%28eval%29%231>
> )  line1  javascript source function () { window.addListener("unload",
> Garbage.empty); if (window.ie) { window.addListener("unload",
> CollectGarbage); } }  failing line
>  javascript call stackReferenceError: "CollectGarbage" is not defined. (
> http://s.bebo.com/script/mootools.v1.11.js#3(eval)#1<http://s.bebo.com/script/mootools.v1.11.js#3%28eval%29%231>)
> at 
> (http://s.bebo.com/script/mootools.v1.11.js#3(eval):1<http://s.bebo.com/script/mootools.v1.11.js#3%28eval%29:1>
> )
> Again, this comes out of our mootools library, which you will see in the
> header of aforementioned bebo site
> InviteJoin page.
>
> On 3/4/08, Marc Guillemot <[EMAIL PROTECTED]> wrote:
> >
> > As you've checked that no js error occurs in "real" browser, this seems
> > to be an issue with WebTest JS support.
> > WebTest JS support (in fact HtmlUnit's one) gets continuously better but
> > is still not 100% as good as it should be.
> >
> > 2 questions:
> > - which WebTest version do you use?
> > - if this is the latest one, do you have a publicly available website
> > where the problem can be reproduced? This would allow to understand and
> > fix it.
> >
> > Cheers,
> > Marc.
> > --
> > Blog: http://mguillem.wordpress.com
> >
> >
> > Ken Hopf wrote:
> > > Hi,
> > >
> > > I'm new to WebTest, so perhaps this is just a silly question on my
> > > part.  I'm running my first WebTest and everything is humming along
> > nicely,
> > > until I try to click a final submit button, at which point I get the
> > > following error:
> > >
> > > javascript errorReferenceError: "CollectGarbage" is not defined.
> > > (http://s.bebo.com/script/mootools.v1.11.js#3(eval)#1<http://s.bebo.com/script/mootools.v1.11.js#3%28eval%29%231>
> > )
> > >
> > > This mootools thing is an Ajax library we use here at Bebo.  Now, when
> > > we run the page interactively, there's no visible problem.  But, it
> > seems
> > > to me that, despite what engineering tells me, this particular error
> > is
> > > not the fault of WebTest but rather of the ajax library (which
> > unfortunately
> > > is compressed).
> > >
> > > I thought that perhaps a WebTest expert might have some insight about
> > > the issue ... eg, whether this really is a problem with WebTest and,
> > > whether it is not, any ideas about dealing with it???   For instance,
> > > does it make sense to anyone that this thing might work fine
> > interactively,
> > > but fail when invoked by WebTest ??   If it is really a problem in the
> > > target application, is there any way to suppress the error aside from
> > > changing the source code ??
> > >
> > > Thanks in advance,
> > > Ken Hopf
> >
> >
> > _______________________________________________
> > WebTest mailing list
> > [email protected]
> > http://lists.canoo.com/mailman/listinfo/webtest
> >
>
>

Reply via email to