You probably wouldn't use the streaming API 20k times/hr. You would open one connection and consume data from it during that hour.
---Mark http://twitter.com/mccv On Sun, Jan 3, 2010 at 12:15 AM, Joel Hughes <joelhug...@gmail.com> wrote: > Hi all, > thanks for your responses. > > John, I did take a look at the stream api but was put off by the big > disclaimer saying it could change very frequently and be down for > extended periods. Also, I was kinda trying to avoid the issue I was > seeing in search where certain tweets were not being indexed - now > perhaps this is because my test tweets unwittingly looked like > duplicates (and not worthy of indexing) but it made me think "ok, is > there a way to avoid this problem - scanning the users timelines > seemed the way to go. > > The streaming API looks interesting though and I get the idea of > having a single connection - which in my case would be a shell based > PHP script dumping the results to (say) a flat file for subsequent > import/processing. I'll need to find some best practice PHP scripts - > phirehose looked interesting. > > Can I check something with rate limits and streaming API? If I have > 20K requests per hour this basically means I can use the streaming API > 20K times per hour? Or is that too simplistic. > > Joel >