Hi all,
After today's team meeting there are a few substantial outstanding
release blockers that prevented the rc from being cut. Most concerning
were the issues related to engage/Red5 and text segmentation. There is
no doubt that our first 1.2 iteration will likely be delayed until early
to mid March, but we should be in a position to cut the RC and test/fix
bug next week. Everyone should be prepared to actively participate in a
major QA effort next week for our RC. I will assign tickets accordingly.
Regarding Red5, we cannot confidently recommend it as a solution for
1.0.x and 1.1 because there is not a strong understanding as to why it
fails to perform adequately in each engage impl. On the same note, the
tests with progressive streaming are promising but have not been
thoroughly tested to conclusively recommend it as an alternative.
Testing a variety of streaming and other delivery solutions will be a
priority for 1.2, but in the interim it's unclear what options the
community can rely on. We need to clearly set expectations.
Please read the following as your involvement may be needed in next steps.
_Capture Issues_
"After recovery from start failure, the agent submits an incorrect
workflow id which results in two workflows in the admin UI. "
Issues: MH-6862, MH-6815
Status: Demoting to critical, but patch is under review.
Next Steps: Micah will merge and test regardless.
"Capture Agent ignores retry config properties and will continue to
retry ingest indefinitely."
Issues: MH-6944
Status: Demoting to critical. If there is a considerable backlog, the
sys admin can ingest them manually and remove them from the capture
agent cache. It's somewhat of an edge case where the retries become
problematic.
Next Steps: Micah will try to address this on Friday.
_Core_
"Ingest temp is never cleaned up if capture agent fails to ingest"
Issues: MH-6936
Status: This hasn't been confirmed nor easily reproducable, and it's
important to know why ingest failed but most importantly why the temp
folder wasn't cleaned up.
Next Steps: Ruediger and others should try to reproduce and provide more
information. Josh/Tobias will need to look at this when they return on
Monday.
"Text segmentation provides inconsistent results"
Issues: MH-6941
Status: The same media provides vastly different results on different
instances of Matterhorn
Next Steps: Josh and Tobias will look into this when they return on Monday
"Ingest via inbox is broken"
Issues: MH-6933
Status: Can anyone reproduce this?
Next Steps: Josh and Tobias will look into this when they return on Monday
"Recordings search doesn't find certain strings"
Issues: MH-6303
Status: Josh fixed but it may be problematic with non-ASCII encodings.
Next Steps: Please test non-english speakers!
_Windows_
"Videosegmentation fails when running matterhorn on Windows"
Issues: MH-6732
Status: Markus M created a fix and will provide Ruben with necessary
binaries to test.
Next Steps: Ruben will test and confirm on Friday
"Windows Cross Compile Failed
Status: downgraded because it cannot be reproduce by multiple developers
_Engage_
There is currently only one issue opened for engage, but based on
today's conversation there are clearly a number of issues that have not
been entered in Jira. Denis and Markus M. will enter these tomorrow,
and Markus K. will work on them on Friday. It's unclear what issues
exist on embed vs. full player, and whether they related solely to Red
5. These examples illustrate the performance issues around streaming.
http://dl.dropbox.com/u/6833781/index.html
_______________________________________________
Matterhorn mailing list
[email protected]
http://lists.opencastproject.org/mailman/listinfo/matterhorn
To unsubscribe please email
[email protected]
_______________________________________________
