Storing the individual tweets in Memcached would work just fine. But there’s no convenient way to retrieve the N most recent tweets. Therefore I’d either need to store the entire list of recent tweets in a single Memcached object, or store individual tweets and store an index of some sort in a different object. Either way is clumsy, and Redis nicely solves this problem with its first-class support for lists.

Curious though why memcached’s support for just strings implies that you’d need to write the entire list of tweets on each new tweet rather than just the latest tweet? Since each tweet has a unique ID couldn’t that serve as the key and on each new tweet you add a new object to memcached with the tweet ID as the key, no?

Regards

My idea: To build a jDownload application, for web. This way rapidshare, megaupload and a lot of others sharing web sites will be able to be downloaded by DNS 321

By itself, neither of these solutions will keep track of the logged in user. If you really need this, you probably need to implement cookies, as these are necessary to keep track of the user session. Otherwise, you may be able to get away with including the basic auth credentials in each request.

I am trying to write a script that uses curl to post data to a tiny webapp that uses authlogic to provide the username/password protection. Does curl support that? I mean, I am guessing one can do a post with username and password (in raw text?) to login, but will curl and rails “remember” that it has been logged in? or do I have to supply username and password for each request that I do?

Thank You!