bug#11037: 24.0.94; gnutls-cli.exe program not found

[Eli Zaretskii <eliz@gnu.org>]

> From: Lars Magne Ingebrigtsen <larsi@gnus.org>
> Date: Thu, 22 Mar 2012 19:33:28 +0100
> Cc: 11037@debbugs.gnu.org
> 
> But I wonder why, er, the following:
> 
> > (signal-process (process-id process) 'SIGALRM)
> 
> > does not work with Win32 Emacs, even with Cygwin also installed.  But
> > one can mimic that with:
> 
> > (call-process "c:\\cygwin\\bin\\kill.exe" nil nil nil
> >                                      "-ALRM" (format "%d" (process-id process)))
> 
> If this is the case, then why not do something along these lines under
> Windows?  That is, define `signal-process' (under Windows) to call the
> external "kill.exe" program (if `signal-process' is broken under
> Windows, as claimed), or fix `signal-process' to do whatever the
> "kill.exe" program does?

Several reasons:

 . kill.exe is part of Cygwin, and won't exist on a Windows machine,
   unless Cygwin is installed.

 . there's no SIGALRM on Windows, so the only way this kill.exe could
   work is if it would send the signal to a Cygwin program (I guess
   gnutls-cli.exe was meant in the message which you cited).

IOW, all this can only work, and as a kludge at that, if a complete
Cygwin installation, including a Cygwin-built GnuTLS, were available
on the target machine.  Which is not really a good way of letting
users of the native Windows build of Emacs to have TLS.  Especially
now, that a built-in TLS is available and working, without any need
for such kludges.