[
https://issues.apache.org/jira/browse/GUACAMOLE-551?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16451745#comment-16451745
]
Michael Jumper commented on GUACAMOLE-551:
------------------------------------------
The warning regarding decreasing timestamps is due to an instruction sequence
like the following:
{code:none}
4.sync,8.82551331;
5.mouse,3.139,3.240,1.0,8.82551281;
{code}
The timestamps above are out of order. This isn't illegal per se, and is likely
unavoidable since two different threads are involved in the graphical content
of the recording (which provides "sync") and the input events from the user
(which provides "mouse"). Each mouse event causes a frame flush through the
same mechanism as "sync", which hits the same timestamp sanity checks,
resulting in a warning.
It's clearer that this is what is happening when looking at the surrounding
"mouse" and "sync" instructions:
{code:none}
5.mouse,3.225,3.239,1.0,8.82551191;
5.mouse,3.224,3.239,1.0,8.82551206;
5.mouse,3.216,3.240,1.0,8.82551216;
5.mouse,3.182,3.240,1.0,8.82551231;
4.sync,8.82551331;
5.mouse,3.139,3.240,1.0,8.82551281;
5.mouse,3.128,3.240,1.0,8.82551332;
5.mouse,3.122,3.240,1.0,8.82551332;
5.mouse,3.119,3.240,1.0,8.82551332;
{code}
If that middle pair of "mouse" and "sync" were swapped, everything becomes in
order. There's an unavoidable race condition there which guacenc's sanity
checks will need to be modified to tolerate.
Overall, I'd say we should be choosing the greater of the last "sync" and the
mouse timestamp when invoking the frame flush due to mouse events, but this
would be a separate issue. I don't think this should be causing noticeable
problems in the quality of the output of guacenc.
> guacenc warning "Layer index out of bounds: -1"
> -----------------------------------------------
>
> Key: GUACAMOLE-551
> URL: https://issues.apache.org/jira/browse/GUACAMOLE-551
> Project: Guacamole
> Issue Type: Bug
> Components: guacamole-server
> Affects Versions: 0.9.14
> Reporter: Manjeeta Raj
> Priority: Minor
> Attachments: MJ.1, MJ.1.m4v, MJ.m4v, command.png, recording.1
>
>
> guacenc instruction throws a warning of "Layer index out of bounds: -1" for
> a recorded file "sudo guacenc /path/to/recording/recordingfilename". This
> issue was raised before too and says its fixed, for which link is attached
> below. Attaching the recorded file as well.
>
> [https://issues.apache.org/jira/browse/GUACAMOLE-307?attachmentSortBy=fileName|http://example.com]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
