On Thu, Mar 5, 2009 at 8:25 AM, Doug Williams <[email protected]> wrote: > > > Lets move to architecture. I can see the following being highly effective: > > [APPLICATION LOGIC] <-> [CACHING LAYER/DATABASE] <-> [TWITTER] > > Where the application logic uses the social graph method to download the > list of friends, then checks with the caching layer if a screen name has > already been resolved. In the case that it has not been previously resolved, > the caching layer accesses the data via a users/show call to Twitter.
That's exactly what I'm doing, but for a slightly different purpose (social network analysis) than the original poster's... but there is one additional wrinkle. Since Twitter users can change their screen names, I'm refreshing the cache every few days for everybody I'm tracking. Come to think of it, I'd love to hear from Twitter how often people really do change their user names. I think I'll post a new thread on that. Nick
