[Feedback] ampache and random play

[Daniel Jones]

> Why don't you just query once for all entries and then do the selection for
> the random playlist on a local copy (of the list)?

It does, but its not really so simple. It can't query for tracks,
because that involves storing information about each track which uses
quite a lot of memory on large collections. It instead queries for
unique ids of tracks.

I'm not sure if such an option exists for ampache (maybe it does), but
it may still be slow. Building a random playlist is easy, but building
a playlists with biases can sometimes involve looking at thousands of
tracks. I'm not sure if remote a collection will be fast enough.

I'm not saying its impossible, it just won't be there in 2.0.