thanks I'll try that, it is possible that they go straight to the login page 
but unlikely so I should be ok.

On 29/mar/2011, at 18.31, Igor Vaynberg wrote:

> simply call session.bind() in requestcycle.onendrequest() which will
> make sure the session is bound on every request.
> 
> you can still have a problem if the login page is the first page hit
> by the user. can that be the case in your application?
> 
> -igor
> 
> 
> On Tue, Mar 29, 2011 at 5:38 AM, Phill <[email protected]> wrote:
>> I'm running into this problem as described in the javadocs for 
>> HttpsRequestCycleProcessor:
>> 
>> "Notes: According to servlet spec a cookie created on an https request is 
>> marked as secure, such cookies are not available for http requests. What 
>> this means is that a session started over https will not be propagated to 
>> further http calls because JSESSIONID cookie will be marked as secure and 
>> not available to http requests. This entails that unless a session is 
>> created and bound on http prior to using an https request any wicket pages 
>> or session values stored in the https session will not be available to 
>> further http requests. If your application requires a http->https->http 
>> interactions (such as the case where only a login page and my account pages 
>> are secure) you must make sure a session is created and stored in the http 
>> request prior to the first http->https redirect."
>> 
>> When my users start a session via the sign-in page protected by 
>> @RequireHttps they are then redirected to a non-ssl but 
>> authorisation-protected page, but as there is no insecure session at that 
>> point they are bounced back to the sign-in page again.
>> 
>> I would really appreciate some suggestions as to how others users deal with 
>> this issue i.e. how could I ensure that a session is created and stored in 
>> the http request prior to the first http->https redirect?
>> 
>> I thought about having a non-ssl protected sign-in page which has an 
>> immediate javascript redirect to the ssl version but it doesn't seem very 
>> elegant.

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to