in the past all recordings have been loaded. At some point this has been
changed to load only active recordings. Now there are params supported
for both forms
all_recordings - all recordings for a broadcast
active_recording - add path of active recordings to broadcast
only_active_recording - show only broadcasts with active recordings
When retrieving individual events, the previous and the following event are determined
according to the current filters to allow navigation between events.
If audio files become inactive because a new file is uploaded, the new
file is marked as active and the old file is marked as not active.
This allows fastly querying events and their active upload file.
Before this change all event's audio files were found and needed to be
filtered afterwards to get a set of unique events.
After the end of a show, the audio recordings are uploaded to a
directory "listen_dir" under a secret name, beginning with the
transmission date "yyyy-mm-dd hh_mm". The directory is available as URL
"listen_url", but must not be listable. In order to gain access to the
recording for 7 days after the show, a check is done to determine
whether the file is available. A random key for the show is stored in
the database if a file matching the date is found in listen_dir. With
this key as a name, a symbolic link to the original audio file is
created. The random name can be used in public documents. If the show
data is read again, the key will be retrieved from the database. File
access can be restricted by the Apache2 handler ListenerAccess.pm. With
it, access to the public symlink is possible for one week.
* remove cache, cache setup is not easy, todays servers do not really
* need it anymore
* add prototypes and fix parameter issues
* suppress redefinition
