Re: Defining HAVE_MOUSE on Cygwin

[Ken Brown]

On 10/7/2012 10:16 AM, Eli Zaretskii wrote:
> > Date: Sun, 07 Oct 2012 08:18:24 -0400
> > From: Ken Brown <address@hidden>
> > CC: address@hidden
> >
> > > > +## Always build with mouse support on Cygwin
> > > > +if test "$opsys" = "cygwin" && test "$window_system" = "none"; then
> > > > +  AC_DEFINE(HAVE_MOUSE, 1, [Define if you have mouse support.])
> > > > +fi
> > > > +
> > > >     AH_TOP([/* GNU Emacs site configuration template file.
> > > >
> > > >     Copyright (C) 1988, 1993-1994, 1999-2002, 2004-2012
> > >
> > > How will this work?  HAVE_MOUSE on a console requires mouse-support
> > > infrastructure.  AFAIK, we only have one such infrastructure: GPM.  Is
> > > that available on Cygwin?  If not, how will the mouse gestures be made
> > > known to Emacs in this build?
> >
> > No, GPM is not available on Cygwin.  But one can still get rudimentary
> > mouse support in Emacs by turning on xterm-mouse-mode, provided the
> > terminal is xterm compatible.  And the standard Cygwin terminal is xterm
> > compatible, i.e., it sends the standard xterm mouse escape sequences.
>
> Cannot Cygwin Emacs be invoked from the cmd window, or from a terminal
> that does not support xterm-mouse sequences?  If it can, what happens
> then if you build emacs-nox that way?

Surprisingly, it actually seems to work when invoked from the cmd window 
(in limited testing).  I don't have a terminal that doesn't support 
xterm-mouse sequences, so I can't test that.

> To tell the truth, I'm a bit nervous about this change.  Code
> conditioned by HAVE_MOUSE makes all kinds of assumptions, which I'm
> not sure are true when xterm-mouse is the agent.  I will have to
> review that code to have an pinion I can defend.
>
> But if that works, my fears notwithstanding, why confine this to
> Cygwin?  If we believe that emacs-nox will always run on a
> mouse-capable terminal, let's turn on HAVE_MOUSE in all Posix builds.

You're right.  If it doesn't work in all Posix builds, then I'm not 
interested in making Cygwin a special case.  I withdraw my original 
proposal.

> P.S.  Any idea why that user couldn't run Emacs compiled with X using
> the -nw switch?

He can do this (and is doing it currently).  He was looking for a more 
"lightweight" solution, that doesn't require installing X.

Ken