I agree it can be complicated. That is likely why it is not a feature as of yet. I think you have to go with a master slave model as I mentioned in my post mentioned above. The sychronization can be automatic or manual as desired by the user. Library archive backups can be performed automatically before changes to allow for regression in case of an error or unwanted / accidental change.
The thing is , none of this is new. Multi user relational databases have been around since I have been working with computers and that is a long time. There just needs to be a command decision made to support it and move on.
Comerical dBs, create timestamps that allow you to roll back to a certain date/time. To be able to do this requires transaction support as every insert/update that is made against the dB is recorded. It would also have to offer internal locking of records during an update. Commercial dBs are designed from the ground up to be multi-user so you can take these features for granted.
MC can save a copy of the dB everytime a change is made and there is support for this currently but this gets unwiedly quite fast. This is strictly a one user option.
As you can appreciate hacking this into the current MC dB is not a trivial task. It would require a shift from its dependance on the filesytem for these features to actually supporting it internally. MC's dB would require a re-write.
At which point JRiver do a cost-benefit analysis and it gets swept under the rug
. This also would imply a certain PRo quality which would be dearer than the current single user version. Would the majority of MC users pay more for this ?
Personally i think this would be a killer feature to have (for the reasons you mentioned) and widen the 2 yr gap between MC & it's competitors. Does that mean they got 2 yrs to implement it
.
Other players have interesting ways of dealing with the same problems.
Music Match for example although quite inferior in many other ways has a seemingly clever way of dealing with this problem. All tag and library information is stored within the file. Even on wav files (some may remember my pleas of many years ago asking for wav file tagging). At any rate when a new instance of Music Match is installed the client just does an import and the library is all there.
You can do this in MC if you override all fields to save to file. But then your backups will take longer. I chose the middle ground where essential tags etc rating are written to the file and the rest is just saved in the library.
The new or updated songs are automatically added to the client through a function called watch folders. Any time a song is added to a watched folder it is automatically imported into the library.
This does not deal well with all the other cool aspects of Media Center's Library like playlists play counts etc. etc., but I think it can be an example of how to deal with the problem of managing or at least adding to the actual music portion of the library.
Thx for clarifying where the watch folder concept orginates from. I have not felt a need for it to date. I have a TO-Listen disk partition where any thing new gets added, then i do an AA and rate it. If i like it it gets shifted to its final destination & imported else it's trashed.
A watch folder would not work in the case as it supposes one or a set of folders and everthing gets dumped in it. I don't want to depend on MC to organise where files are stored, just to present different views of it. I still maintain a genre hiearchy in the file system. So i manually shift an album to the right genre parition before it's imported anyway.
This also has the advantage of reducing file fragmentation as file altering operations are all done in one partition. The only downside is that the playcount is 0 when its imported into MC instead of 1 but that's acceptable to me.
I prefer to explicity import into MC rather than have it automatically do it.
As far as using Remote Desktop to manage the Main Music Server, the JRiver Application does not run as a service and therefore must run as an aplication in a logged on session. As such, to get it to work automatically at boot time, one must auto-login the console session and launch Media Center at startup time.
Not sure i understand this but are you saying its not possible to have MC's server automatically startup after a reboot ?
This may require a 3rd party utility, i vaguely recall it being done that way.
I still think that automated Library: backup (from server), restore (to client), backup (from client) restore (to server) has some short term workaround potential, I just can't figure out how to automate the file naming and execution part.
Sounds quite hairy the way you describe it