I recently made a fundamental change to (setf md-slot-value) in the face of a bug in which a SETF got deferred and I could not think of a way to make the code work without taking out the integrity wrapper in (setf md-slot-value).
That was dumb. The way the slot was being used (setf?! Only in eventhandlers, please.) the slot should not have been a cell. (Specify ":cell nil" in the slot definition.) What I decided to do was go the other way (more Cell-ish) and use a rule to get the slot initialized.
Now I get to go back and take out the dozen extra (with-integrity (:change)...) wrappers I had to add all over the ltktest-ci demo to get that working after the dumb "fix".
Anyway, if you grabbed Cells in the past few days you might want to wait till the smoke clears and grab it again.
And this time I will run the Cells regression test, which I did not think to do last time. <sigh>
kt