Bugs item #1033766, was opened at 2004-09-24 01:43
Message generated for change (Settings changed) made by gregwilkins
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=107322&aid=1033766&group_id=7322
Category: Servlets
Group: None
>Status: Closed
>Resolution: Fixed
Priority: 5
Submitted By: Jeremy Boynes (jboynes)
Assigned to: Greg Wilkins (gregwilkins)
Summary: Request and reponse object are not same after dispatch
Initial Comment:
The request and response object passed to forward or
include must be the same (==) as the argument values of
the target's service method. No wrapping is allowed.
----------------------------------------------------------------------
Comment By: Jeremy Boynes (jboynes)
Date: 2004-10-27 00:10
Message:
Logged In: YES
user_id=378919
Dunno - I was getting 404's from an application that used to
work and when debugging the server saw that the include was
going down this path. I was using whatever snapshot the
build brought in rather than anything local.
If I comment out the filter things work again.
----------------------------------------------------------------------
Comment By: Greg Wilkins (gregwilkins)
Date: 2004-10-26 22:19
Message:
Logged In: YES
user_id=44062
I think something else is going on here. That instanceof
check still works fine because the wrap under mechanism is
implemented as a filter which wraps the original Http
servlet request/response.
So that test still correctly detects the first time that
dispatch is being called (the only time that an unwrapped
request is passed in).
But Alan has just reported 22 other failures mostly due to
dispatching... so something has gone wrong. But as very
little has changed over the last week or two - it may
actually be a
build problem?
I'll make sure the latest snapshot is upto date and then run
Alans tests.
----------------------------------------------------------------------
Comment By: Jeremy Boynes (jboynes)
Date: 2004-10-26 21:36
Message:
Logged In: YES
user_id=378919
I believe the fix for this may have broken dispatch to
resources located underneath WEB-INF.
In WebApplicationHandler we check if the request is being
dispatched by seeing if is it _not_ instanceof
ServletHttpRequest; as the original object is now being
passed though it is, we treat it as an initial request and
return 404 as the path begins with /WEB-INF
----------------------------------------------------------------------
Comment By: Jeremy Boynes (jboynes)
Date: 2004-09-27 02:49
Message:
Logged In: YES
user_id=378919
Thanks
----------------------------------------------------------------------
Comment By: Greg Wilkins (gregwilkins)
Date: 2004-09-27 01:52
Message:
Logged In: YES
user_id=44062
DOH! forced to fix as it is in the spec - for worse or worse!!!!
Created JSR154Filter that has most of the stupid features in
it, namely request listeners and dispatch wrap under. If
the filter is
configured in webdefaults.xml, then you will be J2EE 1.4
compliant - but you will also run slower and have bizarre
stupid behaviour for some code that should be OK.
----------------------------------------------------------------------
Comment By: Greg Wilkins (gregwilkins)
Date: 2004-09-24 23:31
Message:
Logged In: YES
user_id=44062
For more info on this, read:
http://jetty.mortbay.org/jetty/doc/servlet24.html#d0e711
----------------------------------------------------------------------
Comment By: Greg Wilkins (gregwilkins)
Date: 2004-09-24 23:27
Message:
Logged In: YES
user_id=44062
Well to put is simply - DOH!!
I have had this discussion with the EG several times and I
had thought that I had prevailed in the debate. This
"requirement" forces a container to wrap under other user
wrappers and thus violate other parts of the servlet spec.
----------------------------------------------------------------------
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=107322&aid=1033766&group_id=7322
|