http://bugzilla.moblin.org/show_bug.cgi?id=3427 --- Comment #13 from pohly <patrick.ohly(a)intel.com&gt; 2009-07-21 04:04:45 --- (In reply to comment #12) True, at the lowest network level the simulation will differ from the proxy solution. But I think we can ignore this particular aspect by assuming that our transport libraries (libcurl, libsoup) will not get blocked in a real error case and thus our higher-level timeout code is invoked. For libsoup this is guaranteed by the design of the transport (glib event loop), for libcurl we have the 1 minute idle call. Then for the higher levels a dropped message or reply can be simulated: - the wrapper's wait() call always returns ACTIVE even if the underlying transport signals GOT_REPLY - the wrapper must simulate the idle callback when the underlying transport has completed already This is the key difference between the interrupt/resume tests and this issue: if resending works, the current session is never interrupted. This might work with more servers, because as Lukas said, Nokia phones have already used aggressive resending for a while. Right. So we only need to cover the "message sent successfully, reply lost" case. -- Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are watching someone on the CC list of the bug.