diary-mail-entries or mail race condition?
tconnors+xemacs at astro.swin.edu.au
Thu Dec 8 23:12:45 EST 2005
On Fri, 9 Dec 2005, Stephen J. Turnbull wrote:
> >>>>> "Tim" == Tim Connors <tconnors at astro.swin.edu.au> writes:
> Tim> If I run diary-mail-entries, then quit xemacs immediately,
> Tim> then I end up not getting the mail. This didn't used to
> Tim> happen, until we just upgraded the entire system (so timing
> Tim> may be different now -- the mailer might be slower, or who
> Tim> knows...)
> There was a known bug in the stream handling code that under certain
> circumstances dropped data on the floor. I believe this was fixed
> since the date on the version you have, but I can't find the relevant
> ChangeLog (in either 21.4 or 21.5) so I'm not sure.
"Fix: When writing to another process, flush the last chunk, even when the
pipe is blocked." ??
> Another possibility is that the standard mail-handling functions used
> to run the MTA asynchronously, and ignore the return code. We had a
> spate of problems with lost mail, so we changed that to run it
> synchronously. But that means that if you kill xemacs, you'll kill
> the mailer if it's still running.
Checking the system logs show that I don't think the mailer was ever
invoked. I could be missing something though.
> Is diary at the same version? It may have been changed to do things
> in the background.
diary-lib.el just says:
;; Copyright (C) 1989, 1990, 1992, 1993, 1994, 1995 Free Software
;;; (Mostly) Synched up with: FSF 20.5
;;; XEmacs has extra function `diary-countdown'
and is the same as the version I have working at home.
Anyone seeking the "Relativistic Quantum Mechanics" soft option
course, may wish to leave now. -- Intro lecture to RQM
More information about the XEmacs-Beta