See inline

> -----Original Message-----
> From: Matt Sullivan [mailto:[EMAIL PROTECTED]]
> Sent: 13 February 2002 21:15
> To: Cactus Users List
> Subject: Re: setURL question
> 
> Vincent,
> 
> Thank you very much for the clarifications.  It did seem that 8.1
reflects
> the
> how_it_works.html description #5.  I have updated the diagram -
hopefully
> with
> more accuracy ;-)
> 
> Nick,
> 
> Your response really awoke me to the bad assumption I was using for my
> test
> approach.
> 
> It seems that I need to craft a redirector that directs the request at
my
> running application server based on the ServletURL and other request
> parameters.  Am I getting closer?

Not sure I understand that ! By craft do you mean to implement one ? If
so the answer is probably no :-). The goal of Cactus is to do unit
tests, i.e. test directly a given method, without going through the
standard path to reach the method. We're taking a shortcut (the
Redirector) to directly go to your method because, in a given test case
(i.e. a testXXX() method) we're only interested in testing a method (not
the full path). By having several tests you are effectively ensuring
you're testing the full application.

Hope it is more clear.
-Vincent

> 
> Thanks again,
> Matt
> 
> <Copied from Nicholas Lesiecki email 13 Feb 2002 13:36:24 -0700>
> 
> > Hmmm, this is the third or so time someone has misunderstood that
the
> > client side web request does not affect the actual URL cactus calls.
> > Is there anything we can/should do to warn people in the code? Maybe
> > a log entry?
> > "Warning: setURL will not affect the actual http request, which is
> > always determined by the Cactus.properties file."
> >
> > What do people think?
> >
> > Cheers,
> > Nick
> 
> 
> Vincent Massol wrote:
> >
> > Matt,
> >
> > Yes, you're getting closer :-). Everything goes through the
redirector.
> > The setURL is simply there in case you're manipulating the request
URL
> > in your code under test and you're expecting some specific value.
> >
> > WRT to your sequence diagram (a nice one BTW, it would be good to
put it
> > on the cactus web site if you're ok to donate it), here are some
> > comments :
> >
> > 1/ If you're using JspTestCase, the corresponding Cactus redirector
is
> > implemented as a JSP Page (not a servlet although I agree a JSP is a
> > servlet). Same, if you're using FilterTestCase, the corresponding
Cactus
> > redirector is implemented as a Filter (not a servlet although a
Filter
> > can be viewed as a servlet).
> >
> > 2/ Step 3 and 4 are correct. Connection goes to the Redirector.
> >
> > 3/ Step 6 is not completely correct. The redirector does passes to
your
> > test case class the implicit objects (some are wrapped, like the
> > HttpServletRequest, some are not wrapped, like the
HttpServletResponse).
> > But it has nothing to do with the client side WebRequest object.
> >
> > 4/ 8.1 is not correct. The loop is not done on the server side but
on
> > the client side by the JUnit Test Runner.
> >
> > 5/ Step 10 is not "send test results" but "request test result" (no
> > plural, it is done test by test)
> >
> > 6/ Step 11 is not correct. The WebResponse object is created on the
> > client side in step 5 (that is the WebResponse you get in
> > endXXX(WebResponse))
> >
> > Cheers,
> > -Vincent
> >
> 
> < snipped>



--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to