Yes, the behavior you describe would be an improvement. Please request this change at: http://code.google.com/p/twitter-api/issues/entry. Thanks!
On Tue, Nov 4, 2008 at 10:19 AM, Aditya <[EMAIL PROTECTED]> wrote: > > I've not been able to find any documentation for this, and a quick > search of this group also revealed nothing (please feel free to point > me to any existing discussion that I might have missed). > > The quick and dirty: Twitter doesn't like a `since_id` that is not a > part of the timeline it is being sent for. > > The longer and cleaner: Try this - find the status ID of any tweet > (take one from my timeline [http://twitter.com/aditya], I'm pretty > sure you're not following me =P), and make a call to your > `friends_timeline` with this ID as the `since_id`. Twitter will ignore > it completely, and send you the last 20/200 tweets anyway. > > Expected behaviour: The API should return tweets made after the > `since_id` supplied, regardless of whether that status belongs in that > timeline or not. A simple reason is a use case I hit time and time > again: My app stores the status ID of the last tweet fetched, and uses > it for the next call. If I unfollow anyone in between two fetches, and > his/her was the last tweet I received (a pretty common scenario), > Twitter API bonks and sends me tweets I already have. > > What I propose will bring a uniformity to the API call - and maybe > will be easier on Twitter as well (depending on how their fetch-from- > the-database is set up). > -- Alex Payne - API Lead, Twitter, Inc. http://twitter.com/al3x