(synchronised from forgejo)
Currently we regenerate the ids in a non-deterministic way.
This is okay for now but if the api server wants to request a stream for a specific path and has cached some of the information, it will not be useable. We need a shared database of paths and library item id's.
I suggest we create a path-db-item sqlite database which has the direct path mappings between the api server and the transcoding server.
For the transcoding server we might look into foreign data wrapper if they support sqlite.
We also need to ensure the mapping between library -> media item are stable.
(synchronised from forgejo)
Currently we regenerate the ids in a non-deterministic way.
This is okay for now but if the api server wants to request a stream for a specific path and has cached some of the information, it will not be useable. We need a shared database of paths and library item id's.
I suggest we create a path-db-item sqlite database which has the direct path mappings between the api server and the transcoding server.
For the transcoding server we might look into foreign data wrapper if they support sqlite.
We also need to ensure the mapping between library -> media item are stable.