Talk:Cache

What is the rationale for the new format?
The data is rarely a simple string type; it is almost always a serialised structure. Given this, I am not at sure that this is an "improved" format. array('ttl' => $ttl, 'query' => $sql, 'result' => $result)
 * Why not just hoist the serialisation up one level and serialise a context dependent associative array, so in the case of a sql query this would be something like:
 * The  header would seem to imply that you are still thinking of using a php file type for file based acm. Again, I would question this logic because either the LAMP (or equiv) stack supports a proper PHP accelerator cache in which case you would use stored variables (and not stored code), or it doesn't in which case using php incurs the penalties of bytecode conversion without any benefit.  So why not just use a straight data byte stream? :Terrye 17:34, 5 June 2009 (UTC)
 * Terry feel free post a question on Area51 in the 3.0 Discussion forum and I'll happily explain the changes. --Toonarmy 23:48, 5 June 2009 (UTC)
 * I've placed some additional information in the article. --Toonarmy 02:15, 6 June 2009 (UTC)

Expand Article
This can bring unexpected behavior in TTL constraints when using memory based driver which obey the global cache TTL and not fine-grained variable TTL ! --Leodagan (talk) 17:47, 12 March 2016 (UTC)
 * Cache "Key" ($var_name) must begin with an underscore '_' or it will be stored in the global cache... (with memory based driver)