http://bugzilla.moblin.org/show_bug.cgi?id=3427
--- Comment #15 from pohly <patrick.ohly(a)intel.com> 2009-07-24 07:14:23 ---
(In reply to comment #14)
(In reply to comment #13)
I have pushed the patch (in congwu branch, patch for syncevolution and patch
for libsynthesis)
Patrick, could you help review?
Sure. I don't think we are in a hurry with it, because this is a fairly large
change late in the development cycle, so I'd rather release 0.9 first. Then we
can merge it into "master" and later decide where we also back-port it into a
0.9.1 incremental update.
Regarding the code, just some minor comments:
In the SoupTransportAgent it is pretty clear that the callback will happen
roughly after the specified timeout. In the CurlTransportAgent I think you
count seconds by incrementing each time the idle callback is called. This might
be inaccurate (callback not invoked often enough, libcurl changes the call
frequency...). I think it would be better to use gettimeofday() and look at the
delta between start and current time.
I would name and describe the new configuration options a bit differently:
"ResendTimeout",
"The time that the client waits before it assumes that\n"
"a message was lost and sends its own message again.\n"
"In addition, this timeout is used as delay between\n"
"a temporary network error and resending the failed\n"
"message."
"ResendRetries",
"How many times the client will resend a message.\n"
"When the maximum number of retries is exceeded,\n"
"the current session is automatically aborted without\n"
"sending further messages to the server."
Note the second sentence about resending after a temporary network problem. We
haven't discussed that yet and the current code doesn't implement it, does it?
I think it would be a useful reuse of the new error recovery approach. Does
this look useful/feasible?
--
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.