On Feb 15, 2008, at 1:08 PM, Andrew Stromnov wrote:
Sorry, but I'm don't know. This test case - result of my today experiments of using beaker as a standalone caching facility. Beaker unit test doesn't covers this use-case.
Indeed, this is correct. It appears to be a bug in the cache code, its updating the cache retrieval code for new keys when you add settings (this is why key3 uses the createfunc), but its not updating it for existing keys you've retrieved (which is why key1 fails still later).
The proper behavior is that the createfunc should only be used for keys you've asked get_cache for with createfunc in the past. So key3 should fail in addition to key1, since the createfunc was only specified for key2. Is that the behavior you were expecting as well?
Cheers, Ben
smime.p7s
Description: S/MIME cryptographic signature
