On Dec 16, 2005, at 3:18 AM, Justin Erenkrantz wrote:
On Thu, Dec 15, 2005 at 10:12:57PM +0100, Ruediger Pluem wrote:
I think we have to simulate to the client what happened to us on
the backend:
A broken connection.
I mostly agree.
However, Roy's veto is predicated on us not doing anything that
would cause
a hypothetical (*duck*) Waka protocol filter from having the
underlying
connection closed. The point Roy's made is that Waka will have a
'response
abort' indicator. HTTP/1.1 doesn't - therefore, we should abort the
connection for HTTP/1.1.
So, to respect that -1, we need to keep that in mind that we only
force the
dropped connection somehow within the HTTP/1.1 logic. Or, have a
clear
path for a Waka filter chain to not drop the connection - by seeing
the
error bucket and morphing it into a Waka 'response abort' bucket.
-- justin
+1.
The best way I can think of is forcing a client
close by AP_CONN_CLOSE when we see an error bucket.
