bug#46397: 27.1; Cannot delete buffer pointing to a file in a path that includes a file

[Matt Armstrong <matt@rfc20.org>]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Matt Armstrong <gmatta@gmail.com>
>> Cc: 46397@debbugs.gnu.org, eggert@cs.ucla.edu, craven@gmx.net
>> Date: Sun, 14 Feb 2021 14:16:12 -0800
>> 
>> When releasing the lock, I have a less clear opinion but I'm thinking
>> that warnings are better. A warning is still quite intrusive and
>> obvious. Maybe we don't need to decide this part now.
>
> The problem with warnings is that they can go unnoticed, unless
> followed by sit-for.

Hello again Eli, Paul,

I've got a more concrete plan and would like some feedback on it.

While waiting for my signed FSF papers to be acknowledged I have been
investigating how to achieve a prompt in some depth.  I'm coming to the
opinion that issuing a prompt from `unlock-buffer' itself is a bad idea,
but I think prompting from `kill-buffer' is okay.  I could write a whole
essay about why, but instead I'll just propose the following and ask for
your thoughts:

 (a) Modify `kill-buffer' to call `unlock-buffer' sooner, closer to the
     point where it is already running hooks prompting the user.  Handle
     unlock errors there by prompting.  This function already prompts
     the user for the "buffer modified, save anyway" case so a new
     prompt there feels fine.

 (b) Modify Emacs shutdown to handle `unlock-buffer` errors differently
     as well (with `message' or `display-warning').

 (c) Modify `unlock-buffer' to include the buffer name in the errors it
     signals.  It currently only includes the buffer's
     `buffer-true-filename'.  With this the user can more easily find
     and kill the offending buffer.

This approach solves the concrete problems we know about, but doesn't
dramatically change Emacs behavior in other ways.  In other words, Emacs
typically signals errors for file system errors and that stays the same.