[twitter-dev] What are reasonably focused track predicates?
From the streaming API documentation: Reasonably focused track predicates will return all occurrences in the full Firehose stream of public statuses. Overly broad track predicates will cause the output to be periodically limited. After the limitation period expires, all matching statuses will once again be delivered, along with a limit message that enumerates the total number of statuses that have been eliminated from the stream. Limit messages are described in Parsing Responses. My use case is tracking a set of users (follow), and in addition monitoring a list of hashtags (filter). Several questions: a) Does this qualify? b) How many hashtags and/or followers would I have to track to unqualify and start getting rate limited? c) Do follow predicates get a higher priority? If I get rate limited temporarily, what will be limited - the hashtags, the followers, or a random combination of both? d) If one of my hashtags becomes a trending topic, how does that affect the streaming api limit? Thanks! Jonathon Hill http://twitter.com/compwright
Re: [twitter-dev] What are reasonably focused track predicates?
To clarify the taxonomy: The filter method supports three predicate types, represented by the parameters track (keywords), follow (users), and locations (geo). Track and locations are rate limited, but follow is not rate-limited. Follow is never rate limited. If either location or track goes over the threshold, the both turn off. I updated the wiki to be a little clearer. Most trending topics should be fine on the default access rate. But, if you want to ensure that you get all of them, you'll should contact a...@twitter.com for a higher access level. -John Kalucki http://twitter.com/jkalucki Infrastructure, Twitter Inc. On Wed, Mar 17, 2010 at 7:09 AM, Jonathon Hill jhill9...@gmail.com wrote: From the streaming API documentation: Reasonably focused track predicates will return all occurrences in the full Firehose stream of public statuses. Overly broad track predicates will cause the output to be periodically limited. After the limitation period expires, all matching statuses will once again be delivered, along with a limit message that enumerates the total number of statuses that have been eliminated from the stream. Limit messages are described in Parsing Responses. My use case is tracking a set of users (follow), and in addition monitoring a list of hashtags (filter). Several questions: a) Does this qualify? b) How many hashtags and/or followers would I have to track to unqualify and start getting rate limited? c) Do follow predicates get a higher priority? If I get rate limited temporarily, what will be limited - the hashtags, the followers, or a random combination of both? d) If one of my hashtags becomes a trending topic, how does that affect the streaming api limit? Thanks! Jonathon Hill http://twitter.com/compwright
Re: [twitter-dev] What are reasonably focused track predicates?
You will get all tweets up to a certain percentage of total tweet volume. To answer your questions in order a) In the general sense, yes b) It really depends on the specific hashtags and followers. If you have a bunch of trending hashtags followed you may run into limiting c) both. It's based on tweet volume d) It doesn't except that you will likely see increased tweet volume from that hashtag, and therefore a greater likelihood of being limited. My suggestion is to try your full set of predicates. If you see limit messages, remove predicates until you reach a sustainable tweet volume. ---Mark http://twitter.com/mccv On Wed, Mar 17, 2010 at 7:09 AM, Jonathon Hill jhill9...@gmail.com wrote: From the streaming API documentation: Reasonably focused track predicates will return all occurrences in the full Firehose stream of public statuses. Overly broad track predicates will cause the output to be periodically limited. After the limitation period expires, all matching statuses will once again be delivered, along with a limit message that enumerates the total number of statuses that have been eliminated from the stream. Limit messages are described in Parsing Responses. My use case is tracking a set of users (follow), and in addition monitoring a list of hashtags (filter). Several questions: a) Does this qualify? b) How many hashtags and/or followers would I have to track to unqualify and start getting rate limited? c) Do follow predicates get a higher priority? If I get rate limited temporarily, what will be limited - the hashtags, the followers, or a random combination of both? d) If one of my hashtags becomes a trending topic, how does that affect the streaming api limit? Thanks! Jonathon Hill http://twitter.com/compwright
