webchat bugfixes and library updates
I've checked out the webchat project from http://svn.igniterealtime.org/svn/repos/fastpath/webchat/trunk
I fixed some bugs and I updated some libraries.
I attach here the whole modified webchat project and a patch file which shows my modifications to the version I've checked out. The last entry in src/plugin/conf/changelog.html is speaking of version 4.0.0 so I called my modified version 4.1.0
See the attached zip in src/plugin/conf/changelog.html:
4.1.0 – January 3, 2012
- ChatManager session bug fixed
NullPointerException in ChatSession fixed
Updated Smack from age-old version to 3.2.1
Updated DWR from version 1.x to 2.0.8
Ubuntu 10.0.4 (lucid) 64 Bit,
Closing ticket as there has been no comments and a patch was committed. Can reopen if necessary.
Its not clear to me about the status of this ticket, anybody have a comment?
Removing the 'fix version' for all unresolved issues that were scheduled for version 7.8.2. We're releasing this version today - the remaining issues should be rescheduled later.
Am I right to assume that we now can close is a duplicate of this one?
I appear to be running into issue that relate to inactive sessions that take a long time to be removed. Note that "inactivity" here refers to the lack of polling for new messages by the browser client. These checks normally occur every two to three seconds.
The patch as provided here updates the maximum inactivity time from the original one to three minutes. On top of that, the frequency with which this occurs has been slowed down from once per second to once per thirty seconds. Now, on average, it takes three minutes 15 seconds before inactivity causes a chat to be terminated - that is way to long. I've seen that this creates quite some confusion with Spark-sided chat occupants.
I suggest to move the inactivity checks back to once per second, and restore the 60 seconds maximum timeout. Additionally, I'd like to add a mechanism that warns the Spark user of browser inactivity when no polls were detected in the last 10 seconds. This indicates that the browser connection is at least flaky, possibly gone.
What do you guys think of this approach?
Thanks, would you be able to test subversion trunk version out and make sure I didn't screw something up with the commit of this?