On Jul 11, 2014, at 9:12 AM, Traun Leyden <
tle...@couchbase.com> wrote:
> it seems like the if/then check is moot, due to
> _caughtUp = NO;
> always being called before this line.
You’re right. That ‘if’ test is a remnant of some logic that moved into CBLSocketChangeTracker (lines 212-213). In a continuous pull replication, when WebSockets aren’t available, we first get the changes feed using feed=normal, then after that switch to feed=longpoll. (The reason is so the puller can quickly tell that it’s already up to date, instead of waiting indefinitely for the longpoll feed to return something.)
I’ll make the change in the Obj-C repo.
—Jens