Re: controlling echo during mouse-dragging

["David Vanderschel" <DJV4@Austin.RR.com>]

"John Paul Wallington" <jpw@gnu.org> wrote in message
news:<87acyt2c3j.fsf@indigo.shootybangbang.com>...
> "David Vanderschel" <DJV4@Austin.RR.com> writes:
>
> >> > Is there any way I can prevent that uninformative "down-mouse-2-"
> >> > echo from obliterating my infrequent attempts at informing the user?
> >
> >> Try binding `echo-keystrokes' to 0.
> >
> > OK, I tried that.  I made echo-keystrokes buffer-local
> > and set it to 0.
>
> Hm.  I meant `let' binding it when appropriate, but maybe that's
> tricky -- 

No.  A let binding would have been OK also; but this
is a buffer which is read-only to the user and over
which my program exerts full control.  It made just as
much sense to set it once and for all during
initialization.  I never want to see those echoes for
the buffer this program is managing.  I was happy to
learn about the echo-keystrokes variable.

>are you tracking the mouse yourself or binding
>commands to drag events, or something else ?

I have implemented my own track-mouse loop.  (It is
tricky to get right!)

> > So I still have an obliteration problem; and now I do
> > not even know what is causing it.

> Please post some example code, preferably pared-down to essentials,
> that demonstrates the problem.

Actually, I am close to giving up now.  I think I
understand the problem.  The fact is that any message
in the echo area is erased on occurrence of the next
input event.  Normally mouse movement is not treated
as an event, so one is not used to it causing erasure
in the echo area.  However, having invoked
track-mouse, I am saying that mouse movement _is_ an
event, so the echo area gets erased every time the
cursor moves to a new character position.  I do not
see how to get around it.

There is one thing I could hope: namely to get control
of the string that emacs is using, for example, to
display the useless "down-mouse-1-" indication when I
am allowing it.  That is not even a "message" in the
sense that it is logged in the *Messages* buffer -
which is good as far as I am concerned.

Regards,
  David V.