Re: [GNU ELPA] New package: caffeinate-mode

On Sun, Jul 19, 2026 at 11:34 AM Philip Kaludercic <[email protected]>
wrote:

> Lucas Christian <[email protected]> writes:
>
> > Hi all,
> >
> > Long-time Emacs user, first-time contributing a package.
> >
> > One of the more practical additions I am excited about in Emacs 31 is
> > the system-sleep package, which provides a native hook into the
> > operating system's power management APIs.  Long-term, I suspect the
> > potential of system-sleep will come from it being integrated into many
> > Emacs workflows so that the user never has to think about preventing
> > their system from going to sleep while Emacs executes long-running
> > tasks.  Until then, I wrote this package to get system-sleep in front of
> > many users by offering a simple proposition: it can replace the GUI or
> > external command line tool they are likely already using.
> >
> > caffeinate-mode solves the simple and generic case of "I started some
> > long-running task (that doesn't integrate directly with the OS power
> > management APIs), I want to manually tell the system not to sleep so it
> > can complete without interruption".  Users may already be using various
> > utilities for this such as "Caffeine" (GNOME extension or macOS app by
> > the same name), systemd-inhibit (Linux CLI tool), or caffeinate (macOS
> > equivalent).
>
> (Android also has an option by this name.  I didn't know what it does for
> a long time, but apparently this term is becoming more standard.)
>
> >               On previous versions of Emacs, I frequently found myself
> > launching "caffeinate" from M-x async-shell-command for these use cases.
> >
> > Package description:
> >
> >   This package provides two global minor modes that can be toggled on
> >   any time it would be inopportune for the system to sleep:
> >
> >   - caffeinate-mode blocks system idle sleep but allows the display to
> >     sleep.
> >
> >   - display-caffeinate-mode blocks system idle sleep and also keeps the
> >     display active.
>
> >   The modes are mutually exclusive: enabling one automatically disables
> >   the other.  Disabling either mode releases the active power assertion,
> >   allowing the system to resume its normal sleep behavior.
>
> I don't get why you need two separate options for this, can you not have
> one major mode with two values (recall that each minor mode is also a
> user option)?  You can use a prefix argument in an interactive setting
> to force the other mode, depending on which of the two you consider to
> be more interesting.
>

I agree it is a bit wonky to have two minor modes when one would suffice.
I would not make the main menu item mandatory.  I have some stylistic
comments to the code at a general level but if it works for you, it can be
improved over time.

As the author of system-sleep, I specifically didn't add such a mode
because the intention was never to replace O/S utility calls that affect
the entire system but to allow long-running Emacs operations to declare
(via hooks or other in-built policies) they want the system to remain
awake.  e.g., when running M-x compile, one could in
a compilation-mode-hook declare a stay-awake policy.  Separately, an
optional mode-line system sleep indicator could be useful but that seems
separate to your package and would provide a touch more utility across
system-sleep estate that doesn't leverage caffeinate.