bug#67393: 29.1; Slow to open file if autosave exists

[Eli Zaretskii <eliz@gnu.org>]

> From: Ihor Radchenko <yantar92@posteo.net>
> Cc: juri@linkov.net, stefankangas@gmail.com, materus213@gmail.com,
>  67393@debbugs.gnu.org
> Date: Mon, 25 Dec 2023 18:40:08 +0000
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > You do realize that this is contrary to everything we currently do in
> > Emacs, right?  We should the messages that must be acted upon
> > immediately, and use sit-for to make sure the user sees the message
> > and has an opportunity to act upon it.  The message which started this
> > discussion was just like that: it informed the user that an autosave
> > file exists, so the user should consider using it.
> 
> I disagree that sit-for gives an opportunity to act upon the message
> discussed in the bug report: "%s has auto save data; consider \\`M-x
> recover-this-file'".
> 
> Consider that some command opens files one by one in sequence and one of
> these files has auto save data. `after-find-file' will pause that
> command, display the message, block Emacs (not allowing user to do
> anything), and then continue running the command. User has no chance to
> do anything about the auto save recovery until the command is finished
> and also has to wait extra few seconds while Emacs is blocked.
> (This is a real case I encountered with M-x org-agenda)

Such a command, if it existed, should perhaps provide a better
opportunity for the users, like prompt them for whether to recover
from each autosave file before continuing to the next one.

But I was talking about "C-x C-f", the subject of this bug report,
where such a problem doesn't exist.  By talking about a different
command you simply change the subject, which doesn't help to make the
discussion constructive.

> In contrast, what I propose would make sure that the message is
> displayed for at least some period of time after the command finishes.

Imagine a command that needs the user to respond within a short time
interval, after which the message becomes irrelevant, because the
situation changed in a way that the information there is not longer
pertinent.  Like in those mythical Mission Impossible movies: this
message will auto-destruct in 5 seconds.  How will leaving all of
those message on display help in that case?

> Moreover, if multiple files have auto save data, messages about all
> these files will be displayed together without a need to dig into
> *Messages* buffer.

"Dig into *Messages*" is a strange phrase to hear from a veteran Emacs
user.  I'd expect looking in *Messages* to be your second nature.
Some users even have the habit of leaving *Messages* constantly on
display -- there's your "leave messages on display" proposal already
available if someone wants that.