What does 'continuous' mean for PouchDB replicate()?

Question

According to the PouchDB replicate() documentation it's possible to add a continuous option to Pouch.replicate():

If true starts subscribing to changes in the first database and replicating them to the second.

I understand what this means, but how does it work? I've looked at the code, but it's hard to understand how 'subscribing' is accomplished.

Answer 1

It depends on what the first database is.

If it is a remote CouchDB, pouch makes an xhr request to the db/_changes resource for that couchdb (http://wiki.apache.org/couchdb/HTTP_database_API#Changes). It does indeed set feed=longpoll, as Kim mentioned. The longpoll option makes couch wait until there is a change before it sends anything- this way you don't poll over an over only to find there were no changes.

If the first database is a local PouchDB, it listens for "change" events, which are triggered whenever a change is made in the database.

Either way, when the replicator sees a change come in, it replicates it to the second database.

Answer 2

I haven't read the PouchDB code, but the only way to do that is to listen to the _changes feed, probably with feed set to long-polling or continuous.