bug#81578: 32.0.50; completion-eager-display + completion-eager-update breaks cycling + automatic tab completion

> From: Adam Niederer <[email protected]>
> Date: Fri, 7 Aug 2026 20:36:06 -0400
> 
> * Bug One
> 
> Repro Steps:
> 
> - $ mkdir -p a/b/c
> - $ cd a
> - $ emacs -Q --eval "(setq completion-eager-display t)" --eval "(setq
> completion-eager-update t)"
> - Press C-x C-f
> - Press TAB twice
> 
> Expected behavior:
> 
> - The first tab appends "b/" to the minibuffer
> - The second tab appends "c/" to the minibuffer
> - The minibuffer displays "/path/to/a/b/c/"
> 
> Observed behavior:
> 
> - The first tab appends "b/" to the minibuffer
> - The second tab does nothing
> - The minibuffer displays "/path/to/a/b/"
> - Further tab presses do nothing
> 
> I'm not sure if this is intended behavior, but I couldn't find documentation of
> this anywhere if so. It does complicate navigating through singly nested
> directories like e.g. java/com/adamniederer/app/ with these settings enabled.
> 
> I did test one or the other, but both have to be enabled to exhibit this issue
> 
> * Bug Two (probably related)
> 
> Repro Steps:
> 
> - $ mkdir -p a/b
> - $ mkdir -p a/c
> - $ cd a
> - $ emacs -Q --eval "(setq completion-eager-display t)" --eval "(setq
> completion-eager-update t)" --eval "(setq completion-cycle-threshold
> t)"
> - Press C-x C-f
> - Press TAB twice
> 
> Expected behavior:
> 
> - The first tab appends "./" to the minibuffer
> - The second tab cycles, removing "./" and appending "b/"
> - The minibuffer displays "/path/to/a/b/"
> - Further tab presses continue cycling to c/, ../, etc.
> 
> Observed behavior:
> 
> - The first tab appends "b/" to the minibuffer
> - The second tab does nothing
> - The minibuffer displays "/path/to/a/./"
> - Further tab presses do nothing
> 
> Same notes as above, just with respect to cycling

Not sure why you have those expectations.  These two variables seem to
be about updating the *Completions* buffer display and nothing else.

Spencer and Juri, any comments?