"mex" on the GT and input focus

David A. Tristram dat at ORVILLE.NAS.NASA.GOV
Sat Jul 9 04:22:35 AEST 1988


Mark Fichten complains about the key down hack that allows a user to keep
input focus when the mouse leaves his window.  Also, Mark doesn't like
the NeWS paradigm of input to window mouse is over.  Well, the behaviour of
the window manager is programmed using a bunch of PostScript, and is
therefore supposedly easily customizable, hah, but it is not.  In fact,
the PostScript that handles the input events is so crufty and has so many
interdependencies, that it requires a detailed understanding of the entire
body of code to make a modification.  SGI's manuals even recommend hooking
up an ascii terminal if you are going to attempt to program the NeWS server
so that you can recover when you lock it up.  But enough of this.  There is
one more bug in the current input focus paradigm that I would like to see
fixed:

  You hold a mouse button down to keep input focus as the mouse leaves
  a window.  Then you release the button.  The 'mouse-up' event goes to
  the window you are now over, not the original one!  This bugs me.  My
  application continues to track the mouse because it hasn't heard that
  I have released the mouse button.  Having the event go to the original
  window would be an elegant solution, but that's not what happens.

This is the kind of thing that PostScript and NeWS are supposed to let us
get at and fix, but its just not a trivial hack.  Also, I have the feeling
there are only a few people at SGI who understand the mex PostScript and
they are unlikely to read this, or have time to grovel through and fix it.

David Tristram



More information about the Comp.sys.sgi mailing list