Re: itimers and timer-func.el's `run-at-time' clash regarding the current time / zero interval
On Tue, 25 Jan 2005, Stephen J. Turnbull mused:
>>>>> "nix" == nix <nix(a)esperi.org.uk>
writes:
nix> (an exception's thrown during auto-autoloads reading). It
nix> probably breaks other stuff but I wasn't going to go hunting
nix> with an XEmacs that wouldn't start. :)
Urk. Failure of auto-autoloads should _not_ prevent XEmacs from
starting. But that's not what I came to talk about today....
Agreed:)
nix> It seems to me that it's valid programming practice
to have a
nix> fires-now instantly-repeats itimer, because the itimer
nix> function might reset its own repeat period to something else.
OK, but does your patch really implement F-N I-R, or just a very close
approximation on systems that are running sanely?
It implements fires-very-soon (i.e., `now' for all intents and purposes)
instantly-repeats --- although I haven't touched the repeat
behaviour. (All I needed was for fires-very-soon to not give a
seemingly-gratuitous error.)
Fires-very-soon pretty much equals fires-now, because you can't get any
closer without a hard realtime OS. (`Hard Realtime Emacs: for realtime
psychiatry and aircraft control.' An idea whose time will *never*
come...)
What really makes sense here?
I'm not sure. :/ I'd be inclined to say that `repeat instantly' makes
not much sense: if you want that, just use a loop. `Run now' certainly
makes sense and shouldn't be banned, though.
--
`Blish is clearly in love with language. Unfortunately,
language dislikes him intensely.' --- Russ Allbery