Potential mouseinterval() bug/feature

[Byron Stanoszek]

I use the mouseinterval() feature with an argument of 0 to specify the least
minimum time possible to coalesce mouse events into a MOUSE_CLICK event. This
works very well for applications which are only looking for MOUSE_PRESS and
MOUSE_RELEASE events on the local machine.

The moment I try to use this application over a slow network connection,
sometimes both packets get received by ncurses at the same time and the events
are condensed into a MOUSE_CLICK. Having a mousemask set to ignore anything but
Button1 press and release events, the click results in no event sent to the
remote application program.

I propose that a value of 0 for mouseinterval() skips the event-condensing
stage and returns each press/release event as it happens. If this is
unacceptable (i.e. may break existing programs), then a special value of -1 or
a new function should be created to support this feature. Otherwise ncurses has
no ability to turn off mouse clicks for applications that only support a touch-
screen-like behavior.

This patch worked for me to skip condensing events for a value of 0:


--- ncurses-5.2/ncurses/base/lib_getch.c.old    Mon Oct  1 10:45:01 2001
+++ ncurses-5.2/ncurses/base/lib_getch.c        Mon Oct  1 11:18:05 2001
@@ -233,6 +233,8 @@
                if (SP->_mouse_inline(SP))
                    break;
            }
+           if(!SP->_maxclick)
+               break;
        } while
            (ch == KEY_MOUSE
            && (_nc_timed_wait(3, SP->_maxclick, (int *) 0)

Thanks for your time,
 -Byron

-- 
Byron Stanoszek                         Ph: (330) 644-3059
Systems Programmer                      Fax: (330) 644-8110
Commercial Timesharing Inc.             Email: address@hidden