Hi.

* I have a laptop with about X-GB of music
* We share a server with a 250GB connected USB disk (using usb1.1) exporting X-GB of music through smb and nfs (so this is _slow_)
* At work there is X-GB through nfs
* Some other people back home share some music through their windowsboxes

My laptop then have three types of collections

On Tour = local
Work = local+serverX
Home = local+serverY+randomwinmachines

It would be really nice if amarok could handle this, and it acctually consists of several fetaure requests

Several \"collection\" that can be turned on and off
It would be reallly nice to have a \"local\"-collection, a \"work\"-collection and a \"homeserver\" collection that are behaves like one collection when connected but are grayed/hidden when not available. This could be achived by turning them on and of manually or do some filesystem checks to see if they are available.

This also makes it possible to maybe add a file to your local collection from one of the other collections even when offline. So If you could browse the total of all collections, choose witch songs you want and when that collection becomes available it will copy it over. (I belive itunes does something like this to ipod-mini)


1. Handle smb://, and maybe any other kio_ plugins
This would be really nice to use with random machines out there and also makes it unneccesary to smbmount a share to make it work more flawlessly like \"remote connections\" in konqueror.


Handling fast updates on non local files
Amarok uses fam on local files (and maybe nfs?) but probably searches the full directory if fam is not running or on a remote smb or similar-disk

I once implemented a music-server called musicbase. It had a postgress or sqlite backend and just used timestamps on directories. It replicated the \"compilation\"-disk structure into a database + all the tag dataes etc, so every file had a pointer to it\'s parent. and also to it\'s root \"compilation\"-directory. This way i could easly just show the available parts of the compilation, and by storing the mtime of the directories and files, i could do some nice things when updating the compilation

Start at rootnode

If directory is new, add directory
if directory mtime is updated, this means that a child file or directory has been updated
Check every file and directory agains mtime in database, update if newer, add if new, delete if gone, recurse into directories.

By using this on a collection on about 146GB of music i used over 2 hours to run the first time, but not more than 10 minutes the second time (even with some added files)

So two gaines out of this. I could easily browse the catalog structure on extremly slow remote disks (even much better than my 4200rpm laptop disk), I had everything easily searchable like in amarok, and bringing a collection up to sync only takes about 10 minutes (for 146GB on slow disk) only a few sec on fast disks. I also could hide some parts of the collection by using filtering on the root note (every \"collection\" had it\'s own root node (\"/local/music\" \"/mnt/remote/machine/music\" etc)


For reference, the musicbase code is here:

http://cvs.sourceforge.net/viewcvs.py/m ... musicbase/

Audun Ytterdal wrote:

1. Handle smb://, and maybe any other kio_ plugins
This would be really nice to use with random machines out there and also makes it unneccesary to smbmount a share to make it work more flawlessly like "remote connections" in konqueror.

Your idea about several databases is a good one, and I would be happy if it was implanted.
About kio_plugins, they already work. But
1; You can not add them do the database, and
2; You need to use the right playback engine to playback them. Gstreamer supports the kio_plugins, so if you use it and use the filebrowser in amarok(or drag and drop for konqueror) and go to e.g smb://computer/Share and drag those to the playlist they should playback just fine.

mortiferus wrote:

Your idea about several databases is a good one, and I would be happy if it was implanted.

It\'s not several databases, You just have to add a extra field in the filestable that refers to the "mountpoint" of that collection so you can easily filter away the ones that are not available etc...

About kio_plugins, they already work. But
1; You can not add them do the database, and

Is this because not alle backend support files like that? If the engine supports it, why not?

2; You need to use the right playback engine to playback them. Gstreamer supports the kio_plugins, so if you use it and use the filebrowser in amarok(or drag and drop for konqueror) and go to e.g smb://computer/Share and drag those to the playlist they should playback just fine.

Yep, I know. But it would be much better to acctually be able to add them to your collection.

I\'d really love to see this feature implemented in amarok, as i don\'t know any other player that does it (neither windows or linux, where amarok rocks all the others).
I have 2 pcs at home, one is a laptop and i\'d like to be able to listen to the music mounted with samba and to be considered into the collection just when i\'m at home using my local network. In any other situation, my library should have only the local files.
Pleeeeeeaseee, make that feature