Re: [HtmlUnit] Redirects

[Thomas B. <th...@cy...>]

Hello again,=20

thanks a lot for your quick answer.=20

Am Montag, 23. September 2002 15:57 schrieb Mike Bowler:
> > In HtmlUnit the submit of the
> > form-parameters via is POST is correct. But when it tries to
> > follow the redirect, it somehow converts the POST parameters
> > into GET parameters and appends them to the URL-string. That's
> > not right behavior and our application can't handle that.
> > Is there a possibility to switch that off or does anyone know a
> > workaround?
>
> You can disable automatic redirection handling with
> WebClient.setRedirectEnabled(false).

Thats not exactly what I had in mind ;)

> It's certainly possible that I've made a mistake in the redirection
> handling but as far as I know, it is correct according to the HTTP/1.1
> specification.
>
> http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html#sec10.3
>
> If you think that HtmlUnit is handling this incorrectly then please
> open a bug report.  I'll need to know the exact status code being
> returned from the server as the redirection logic is different for
> different status codes.
>
> Since you mentioned the method switching from post to get, I assume
> that the status code you got was either a 302 or a 303.

So after I got your mail I checked it all a little bit and found out the=20
following: The given status code was a 302 and the HTTP/1.1 specification=
=20
doesn't really mention what to do with POST-Request-Parameters on a 302, =
does=20
it? As far as I understand, it says to keep the original Request-URI for=20
future requests, not convert parameters for the redirect-URI. But maybe I=
'm=20
wrong with this, after reading the specifications a few times I was prett=
y=20
puzzled how to understand it at all ;)
However, I tried something different. I compared the behavior of an inter=
net=20
browser (Mozilla) with the behavior of HtmlUnit. It is in fact so, that t=
he=20
browser doesnt convert the original POST-Parameters into GET-Parameters w=
hile=20
doing the redirect and HtmlUnit does. I tried this out with two simple=20
PHP-scripts. The first one contains a form, which points to itself (PHP_S=
ELF=20
as action), but redirects to the second when the variable for the=20
submit-button is set. The second one just prints out the value of the=20
form-datas if accessible, like the vaue of the submit-button for instance=
=2E So=20
with browser I don't get any output on the second page, like I get with=20
HtmlUnit.=20
Well, as I said, I'm not sure how it really is specified. But however it =
is,=20
HtmlUnit does here something uncommon. The most user agents (IE, NS,=20
Opera...) handle it like described above and different from the HtmlUnit-=
way.

It's up to you if it's a bug in HtmlUnit or actually in all other clients=
=2E In=20
case its not a bug, I'd like to see a possibility to configure browser-li=
ke=20
behavior on redirects.

Thanks for your help anyways.=20
Kind regards,
Thomas
--=20
Thomas Bartz
th...@cy...
G=E4rtnerstr. 28, 10245 Berlin; Tel.: 030 29005717
http://www.cyainhell.de