This page is of historic interest only.
Details of sessions/events to be recorded, the recordings made, converted files, and team members, should be stored in a database. It should be possible to pull session information from COMAS or other conference software; in fact if that is run on-site then we can probably use a view onto its table of sessions. However, It must be possible to make a recording without access to the database and to add it later.
There should possibly be a central daemon deciding what encoding work is to be done, though that may not be necessary.
 Recording etc.
Each recording/encoding/streaming system should have a daemon that:
* starts recording at the request of a client (which would run on the grabbing machine) * adds new recordings to the database * does off-line post-processing and encoding work when not recording * post-processing can be quite complex, and some of it may need to be left until after the conference * adds finished encodings (whether successful or not) to the database * provides a continuous stream, showing a live recording when available and an information screen at other times * and for audio...? guess we could use a speech synthesiser...
It should be possible to use the same software with the streaming and recording disabled to provide extra encoding capacity.
New recordings can be live or grabbed from tape. We should tag them with the time they were made, based on the timestamp from the DV but overridable if necessary. Recordings from tape shouldn't be streamed.
 User interface
After starting a recording the client should prompt for metadata. For live recordings this should default to information about the current session according to the database, but we must allow for changes in case of schedule changes and overruns. I would prefer to use a command-line interface on the client so I'm not sure how this form-filling should be done...
There should be a web interface where volunteers can:
* assign themselves work on the upcoming sessions * correct metadata about recordings * find unchecked files, claim them, download them, report on them * claims should time out in case people forget to finish the job * we should define post-processing filters in the database and allow reports to specify which of these should be used * see a summary of what work they have agreed to do
Possibly there should be a role for an administrator who can assign work to other people (resulting in mail to them?).