I have a GreaseMonkey script ( http://userscripts.org/scripts/show/64286 ) which displays the "Follows in Common" and the "Follow Rank" for a given users profile page using the Friends and Follows api requests.
I've implement cursor style json request to the api, but have several concerns around the elimination of the legacy Social Graph API data format particularly with request to getting very large sets of ids ,or what I come to call "the Oprah problem" (but could just as easily be called "the Kutcher problem" Doing the legacy style page requests did work for this problem this is always resulted in a "badly formed json document" error. However the cursor style json request it not much of a improvement since the cursor request only results in (on average) ~5,000 requests which given her 2,937,081 followers means It take 588 requests to get the ids set. That would take > 6 minutes if - remember this is a GreaseMonkey script running in a firefox browsers - it did not hit the 150 request throttle limit!! (I do cache the final result, I may have to cache the incremental results and ask users to try again in a hour, and then try again and again (!!) ) White listing won't solve this particular case of problems. Beyond just increasing the request throttle limit, which would allow completion in my case, an additional change would be to increase the size of the cursor set to 50,000 to 200,000 which would go a big way to resolve some or most of the performance issues raised by others. Any other options, comments or solutions? Ian Irving
