Hmm. Where does the time go?
The ANSI CL test suite is now at over 18,000 tests. Right now I'm testing interactions between various kinds of unusual vectors and the builtin string/sequence functions. This is mostly to support crhodes in his effort to add unicode support to SBCL, but it's already found at least one five year old bug in CLISP (REVERSE didn't work on displaced vectors).
After that I'm going back to finishing the FORMAT tests. I'm still aiming for about 20K tests for the 'finished' test suite.