Auto cleanup as data grows about ground-db

[raix]

you could in theory add some offline metrics like size/usage/age of documents, then have a sort function using weights to figure out the most important documents. I've done this in the past and wanted this in gdb at some point. It's something to think about.

[ramezrafla]

Thanks @raix

I had an idea to make sure we keep data well sync-ed with server. The problem is with all the subscriptions, it's tricky running a keep as you may lose data that is still on server but not subscribed to yet.

We could have a server method to which we send all the ids currently in gdb and it returns the ones that are valid, and we would run on the client a keepIds (a clone of keep but that depends on id's).

Thoughts? If you are willing to accept a PR, I could work on it. It could be a simple client method called sync() that would take care of everything.