Re: Mouse events in the margin using an xterm

[Nick Roberts]

 > > More accurately, t-mouse.el uses start-process to run the `mev'
 > > program, which in turn communicates with `gpm', and feeds any mouse
 > > events to the Emacs event queue through the process-filter.
 > 
 > Hmm... I guess that Emacs pre-CVS can only connect to a TCP socket.
 > With the new socket primitives maybe we can get rid of the
 > `mev' step?

I think Eli is saying get rid of the "gpm" step to and do everything in Emacs.
I was looking at writing lisp primitives to get the information for mouse
events out of Emacs to use in t-mouse.el. Clearly it would be a lot better to
to do it all internally, just as it must currently be done in X11. Evidently,
there is also a program call rmev, that works on an xterm (or rxvt) so
presumably xt-mouse.el could be replaced too. However, I am not familiar with
the low-level activity involved in such a project.

I think this would be a good activity to put in TODO.

The current modes work quite well but they have (at least) the following
shortcomings:

1) You can only drag the mode line when you click on any part of it with
   mouse-1.

2) You can only drag the header line when you click on any part of it with
   mouse-1.

3) Mouse click events in the margin can't access the extra content that Kim
   has given them. In particular this means that you can't toggle breakpoints
   there.

4) Multiple clicks don't work (xterm.el)

5) Mouse-2 doesn't work with a two-button mouse (simulated by pressing
   mouse-1 and mouse-3 together). (t-mouse.el)

Nick