From: Richard M. Stallman <rms@gnu.org>
Date: Mon, 4 Sep 2006 15:07:32 +0000 (+0000)
Subject: (Idle Timers): Explain why timer functions should not
X-Git-Tag: emacs-pretest-22.0.90~769
X-Git-Url: http://git.eshelyaron.com/gitweb/?a=commitdiff_plain;h=0a369706ff1743a115dd39648df6487ebe2f5025;p=emacs.git

(Idle Timers): Explain why timer functions should not
loop until (input-pending-p).
---

diff --git a/lispref/os.texi b/lispref/os.texi
index f6682548e5b..edf5833bf42 100644
--- a/lispref/os.texi
+++ b/lispref/os.texi
@@ -1578,6 +1578,25 @@ Here's an example:
 @end smallexample
 @end defun
 
+  Some idle timer functions in user Lisp packages have a loop that
+does a certain amount of processing each time around, and exits when
+@code{(input-pending-p)} is non-@code{nil}.  That approach seems very
+natural but has two problems:
+
+@itemize
+@item
+It blocks out all process output (since Emacs accepts process output
+only while waiting).
+
+@item
+It blocks out any idle timers that ought to run during that time.
+@end itemize
+
+@noindent
+To avoid these problems, don't use that technique.  Instead, write
+such idle timers to reschedule themselves after a brief pause, using
+the method in the @code{timer-function} example above.
+
 @node Terminal Input
 @section Terminal Input
 @cindex terminal input