Hi Taylor, I am using [Tweetr APIs] for my application. I want to get user's public feeds. However, i am facing a GET request rate limit problem. I am wondering even if i make one or two request, it is giving me 400 error [rate-limit reached]. If you can explain me the reason or any workaround for this problem, it will be great....
Thanks & regards, Bhushan On Wed, Jun 9, 2010 at 12:41 AM, Taylor Singletary < taylorsinglet...@twitter.com> wrote: > Hi Ed, > > I think you're doing the best that you can do to be fault tolerant in this > case. We generally recommend exponential back-off in the face of continued > error.. perhaps waiting 5 seconds before retrying after the first failed > request, then widening to a longer duration, and so on with each subsequent > error. It is recommended that you implement this kind of behavior, as in > times of high error rates, those applications that are ignoring error codes > and retrying the same requests most aggressively are candidates for > temporary blacklisting (to relieve the unproductive stress on the system as > it recovers from error states). > > Your normal operation behavior also seems to be the correct one to utilize. > Though if you want to implement additional waiting time, that's up to you. > Dynamically handling rate limiting is a good idea, as one shouldn't really > expect the rate limiting to a constant function/rate (though they generally > are today). > > Taylor Singletary > Developer Advocate, Twitter > http://twitter.com/episod > > > > On Tue, Jun 8, 2010 at 11:36 AM, M. Edward (Ed) Borasky < > zn...@borasky-research.net> wrote: > >> I have a Perl script that downloads historical tweets using the >> "user_timeline" REST API call. I'm running into 503 - "Bad Gateway" - >> "Twitter / Over capacity" errors when I run it. Questions: >> >> 1. When I run into an error, I'm waiting 45 seconds before retrying. >> Should I wait longer? Is there a shorter recommended wait time after an >> "Over capacity" error? Do I need to wait at all? >> >> 2. In normal operation, I'm using the returned rate limit header >> information to pace the request rate so that I never run out of calls. This >> can generate a call as soon as I've completed processing of the previous >> data. Should I insert a non-zero wait time here? I've tested explicit wait >> times as high as 20 seconds here and they don't seem to be reducing the >> incidence of "Over capacity" errors. >> >> >