https://bz.apache.org/bugzilla/show_bug.cgi?id=64762
--- Comment #11 from Ralf Hauser <hau...@acm.org> ---
The problem still persists, albeit we were able to almost make it disappear
We no longer wait a limited amount of sleep iterations, but "forever" (provided
the browser declares "content-length").
Reasoning: we are on a quick fiber network and most of the affected users
normally are on arbitrary (slow) home-office networks and only experience it
intermittently.
So if we don't wait forever, they will see an error, so we better wait until
their browser-timeout anyway and hopefully get the last expected packets before
that happens.
Another strange, but also locally reproduceable phenomenon is that normally
HttpsServletRequest request.getAttribute("javax.servlet.forward.request_uri")
is not null, but under certain circumstances it is null.
In this case request.getRequestURI() which eventually gets to
org.apache.coyote.Request.requestURI() still has the value sought.
Further analysis show that if HttpsServletRequest is instantiated by
org.apache.catalina.core.ApplicationHttpRequest the attribute has a value but
if it is rg.apache.catalina.connector.Request (inside
org.apache.catalina.connector.RequestFacade), all its source of attributes are
empty (specialAttributes, attributes, coyoteRequest.getAttribute()) i.e. null
--
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
