Minimal PSR-16 cache-implementation
This library provides a minimal PSR-16 cache-implementation backed by simple file-system storage.
This can be used to provide working, lightweight bootstrapping when you want to ship a project that works out of the box, but doesn't depend on an awesome, full-blown caching-framework.
Files are stored in a specified cache-folder, with two levels of sub-folders to avoid file-system limitations on the number of files per folder. (This will probably work okay for entry-numbers in the tens of thousands - if you're storing cache-entries in the millions, you should not be using a file-based cache.)
To reduce storage overhead and speed up expiration time-checks, the file modification time will be set in the future. (The file creation timestamp will reflect the time the file was actually created.)
Please refer to the PSR-16 spec for the API description.
In a production setting, consider specifying appropriate
$file_mode constructor-arguments for
your hosting environment - the defaults are a typical choice, but you may be able to tighten permissions on your
system, if needed.
Because this is a file-based cache, you do need to think about garbage-collection as it relates to your use-case.
This cache-implementation does not do any automatic garbage-collection on-the-fly, because this would periodically block a user-request, and garbage-collection across a file-system isn't very fast.
A public method
cleanExpired() will flush expired entries - depending on your use-case, consider these options:
For cache-entries with non-dynamic keys (e.g. based on primary keys, URLs, etc. of user-managed data) you likely don't need garbage-collection. Manually clearing the folder once a year or so might suffice.
For cache-entries with dynamic keys (such as Session IDs, or other random or pseudo-random keys) you should set up a cron-job to call the
cleanExpired()method periodically, say, once per day.
For cache-entries with dynamic keys in the millions, as mentioned, you probably don't want a file-based cache.