For folks who use JMeter to load test wicket apps and are having problems
with 404s on redirect URLs containing '/..': I posted a patch to JMeter
trunk that fixes the problem by simulating behavior of major browsers -
collapsing the URL before issuing the successive GET:

https://issues.apache.org/bugzilla/show_bug.cgi?id=49083

apply to JMeter trunk

On Thu, Apr 8, 2010 at 12:06 PM, Nikita Tovstoles <
[email protected]> wrote:

> ahh, good point. I was looking at this:
>
> http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.30
>
>  <http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.30>Are you
> suggesting it's Tomcat's responsibility to either:
>
>    - collapse the '/../' when generating the absolute URL for the Location
>    response header OR
>    - deal with those URLs when parsing a request?
>
>
> -nikita
>
> On Thu, Apr 8, 2010 at 8:35 AM, Nikita Tovstoles <
> [email protected]> wrote:
>
>> I am working on a patch to JMeter to rewrite absolute URLs containing
>> '/../' before issuing GETs and have an observation and a question. The
>> problem is described here:
>> http://markmail.org/thread/3ci5hayvqyjhi2ud
>>
>> FWIW this was discussed before and there's WICKET-2728 and related JIRAs.
>> My understanding is that the reason wicket chose to break the RFC and issue
>> relative URL redirects is to accommodate a 'real world' case. I can
>> understand that.
>>
>> Unfortunately, Response.encodeRedirectURL in Tomcat will rewrite
>> redirectURL to absolute anyway, thus returning Location as absolute yet
>> containing '/../'  that - if replayed as is - will result in a 404. (see
>> toAbsolute() in
>> http://kickjava.com/src/org/apache/catalina/connector/Response.java.htm).
>> So the benefit derived from breaking RFC seems to be obviated, no?
>>
>> So, given that Tomcat is widely used container, what was the logic in
>> trading off RFC compliance? Just trying to understand the philosophy.
>>
>> thanks
>>
>> -nikita
>>
>>
>>
>
>

Reply via email to