I ran across an article on "How to build a simple caching system, with PHP" on PHPit today. Overall, it's a fairly decent article, and uses some good principles (using the output buffer to capture content, using a callback to grab the captured content). There are a few minor improvements I'd make, however.

There are some definite areas I'd change. First off, die() doesn't make sense to me as a way to abort execution if content is found. exit(0) makes more sense — it indicates that execution was successful.

Also, it's often useful to have the ability to clear the cache for a given page. Adding a check for a $_GET or $SERVER['PATH_INFO'] element could accomodate that.

The article says to include the file with the caching functions on every page. I have a couple issues with that:

1. Too easy to forget to include it.
2. Namespacing — using function in the global area could conflict with other user defined functions.

To solve (1), use an auto_prepend_file directive. You can do this in a .htaccess file (php_value "auto_prepend_file" "/path/to/cache.php"), in the httpd.conf, or directly in your php.ini; in either case, it only needs to be defined once, and you never have to worry about it in your scripts. If there are scripts you never want to cache, you can override the value in individual .htaccess directives, or setup a hash-lookup in the caching routines to skip such scripts.

To solve (2), wrap the functions into a class. You could still call items statically (Cache::get_url(), etc.). This would also allow you to define the hash lookup as noted above — simply place it in a static property.

Finally, this has all been done before. PEAR::Cache_Lite offers all of this functionality (minus the routines to create unique identifiers per page), and a little more — efficiently, even. The only difference I've seen in practice is that when using PEAR::Cache_Lite, I had to do an auto_append_file as well to stop the output buffering.

Overall, a nice article and introduction to page caching.