* Added session auth to the Media API (if you're logged in)
* Started reworking our Plupload interaction with the server to be less
awkward.
* Basic uploading works with the Add Media page again, but messages
aren't dispatched to airtime_analyzer yet (coming next...)
* Fixed a race condition in pypo which could kill playout. Input kick
events and non-file/webstream events could result in an exception
being thrown. Fixed the logic error behind this, caught the exception
just in case, and added some code to prevent uncaught pypopush
exceptions from terminating the thread.
* Could prevent playout breakage in similar situations
* Added MetadataAnalyzer unit tests and test data
* Improved debug logging and squashed pika logging
* Implemented file moving
* Extract the track number/total
* Fixed mapping of mutagen to Airtime fields in a few spots. The mapping
matches the DB column names now.
* Fixed the bin/airtime_analyzer binary
* Started work on PluploadController to make it work with the new File
API
* Basic HTTP reporting back to the File API works (PUT)
* Use the database table names as JSON field names.
* Fixed result returning bug in message_listener.py
* Fixed API key verification to adhere with the HTTP Basic Auth spec
* Added StatusReporter class to make HTTP requests back to Airtime
* Handle even more errors now
* Added proper logging and log rotation
* Added --debug flag for increased logging.
Disabled scheduled content into instances that have already played out
Removed setting the show's start date to the first repeating instance's start date
because we already fixed that in CC-5697
* Finished the skeleton of the airtime_analyzer service.
* Basic round-robin, reliable AMQP messaging works.
* Using multiprocess arch so the daemon survives analyzer crashes and
avoids failures propagating to other nodes.
* Basic metadata extractor using Mutagen is done.
* HTTP requests to the File API to are next to come...
The show content was getting recalculated twice sometimes.
So if the difference between the original and now show start
date-time and was 2 hours, the scheduled content would have
a difference of 4 hours
The problem was that we weren't updating the first show's start time if deleting
a repeat show day changes it. This affected the difference between the original
show start time and the new one.
Fixed a problem where we can't store the original cc_show_day data
because none can exist at some points. If the user changes the only
repeat day, the cc_show_day rule is deleted first then recreated
Albert Santoni
localizer
Naomi
drigato