On Thu, 05 Oct 2006 05:39:58 +0300, Eugene Paskevich
<[EMAIL PROTECTED]> wrote:
Issue 1:
It happens sometimes that typing notification state is left for hours
(lost/unparsed packet?) when remote user is surely not typing... It is
stripped off when the next "stopped typing" is received or when the user
goes offline (I believe).
Unless protocol has some way of verification of such state, we could
strip it off irregardless in 5 minutes or so after the moment we have
received last typing notification. (Yes, I understand, that this
involves a new timer, but what else could be done here?)
It last I have found a way to reproduce this stale notification...
Scenario:
1) I open up a message window.
2) Start constantly typing.
3) Stop typing.
4) Close the window just before it is time to send out the notification
that I've stopped typing.
That's it. Notification stays forever active until client sends a new stop
or goes offline.
'Close window' event should trigger 'typing stop' as well.
--
Eugene Paskevich | *==)----------- | Plug me into
[EMAIL PROTECTED] | -----------(==* | The Matrix
