Thank you guys for your insight. I was in a rush so I picked simple search but I'll revert it to stream. Not quite clear what do you mean by saying to revert stream every hour but I guess it's in the docs.
On Aug 13, 1:58 am, Mark McBride <mmcbr...@twitter.com> wrote: > In both cases it's still probably best to use streaming. You don't > want to connect to often, but once an hour should be totally fine. > > ---Mark > > http://twitter.com/mccv > > On Tue, Aug 10, 2010 at 6:34 AM, Tom van der Woerdt <i...@tvdw.eu> wrote: > > > On 8/10/10 12:58 PM, bitstream wrote: > >> Hi all, > >> I've been reading api docs lately but still can't figure it out what > >> will be the best approach when searching for hashtags. > >> streaming > >> I know that streaming api support statuses/filter where I can declare > >> 'track'. It's possible to use statuses/filter and add a track on > >> '%23hashtag' ? > > >> search > >> Or use a simple approach by calling search api and parse response from > >> something like this:http://search.twitter.com/search.json?q=%23hashtag > > > My opinion: > > > It depends. If you want to track a lot of keywords, you should use > > streaming. If you track only one keyword, then both are an option, > > depending on the amount of tweets for the hashtag. If you have a lot of > > keywords but they vary (for example, when users can add/remove hashtags) > > then you should consider a combination of both, where you reset the > > stream every hour and update it with new hashtags, and use the REST API > > for the hashtags that get added in the hour. After all, you don't want > > to reconnect too often. > > > Tom
