Caching to disk, at least for ports, is now implemented. I didn’t plan to moved a new feature into production on the day before I leave for a conference, but it happened.
If you notice that the website has not been updated after a port has been committed, please let me know. That will indicate a problem with the cache removal process. The code that processes commits should be removing cache entries whenever a port is touched by a commit.
At present, a commit against a master port will not remove the slave ports from the cache. That is pretty straight forward to fix. Perhaps I will do that while away at USENIX ATC in Boston. I leave tomorrow morning for the 8 (or so) hour drive. I hope to be on the road by 7 am.