According to facts stated in [Richard's post #19](https://sqlite.org/forum/forumpost/e78b285637), which can be independently seen in [the project VCS timeline](https://sqlite.org/src/timeline), that snapshot was made at nearly the same time according to its filename timestamping and incorporates a change responsive to the issue raised in this thread. While I have no reason to doubt Richard's word on this, I can also see that no other recent checkins [a], since v3.32.2 which reproduces the misbehavior, do anything to address it. [a. I take the checkin comments at *their* word, without deeper examination. ] Hence, it is fair to conclude that the change fixing the misbehavior was instigated and motivated by Richard's reading of this thread.