> These were not idle questions.  I wrote the patch, and I had to know
> what behavior is correct.

Ah, sorry. I assumed you were going to submit a separate patch to fix the
unconsumed_tail issues.

> Here's the patch.  It fixes potential memory bug (unconsumed_tail sets
> to NULL in case of out of memory), resets the unconsumed_tail to b''
> after EOF, updates unconsumed_tail and unused_data in flush().

Did you perhaps forget to attach the patch? The only ones I see are those
that you uploaded last week.