Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!ames!elroy!mahendo!jplgodo!wlbr!scgvaxd!trwrb!aero!venera.isi.edu!raveling
From: raveling@vaxa.isi.edu (Paul Raveling)
Newsgroups: comp.windows.x
Subject: Re: X Conference notes
Message-ID: <4658@venera.isi.edu>
Date: 29 Jan 88 18:21:45 GMT
References: <8801221812.AA19881@> <181@granite.dec.com>
Sender: daemon@venera.isi.edu
Reply-To: raveling@vaxa.isi.edu.UUCP (Paul Raveling)
Organization: Information Sciences Institute
Lines: 52

In article <181@granite.dec.com> rost@granite.dec.com (Randi Rost) writes:
>
>To anyone who read Paul Raveling's trip report:
>
>I just wanted to clear up a couple of misimpressions about the PEX (3D)
>extension for X.  The first potentially misleading statement in the
>report is:
>
>    "A PHIGS window is EXCLUSIVELY PHIGS; external X access is not
>    practical."
>
>If I'm reading this correctly, the statement is untrue.  ...

	If this is substantially misleading, I apologize for the
	mis-statement.  Both the talks and the notes necessarily
	condensed lots of material into perhaps excessively brief
	remarks.

	Also, I was doing a little "reading between the lines" by
	relating the PEX architecture to issues (a euphemism for
	problems) we encountered in work directed toward supplying
	an integrated user interface for existing software.  The
	most relevant issues arose in connection with our Geographic
	Display Agent, and at first sight it would appear that
	variants of these would affect mixing separate X and PEX
	i/o in the same window.

	Briefly, some of the biggest problems in our X10.4-based
	experience were ...

	--  Inability to direct particular X events for a given
	    window to different processes, each with its own
	    connection to the display server.
	
	--  Use of nonlinear world coordinates in a map window
	    showing a Mercator projection, with only linear
	    coordinates available to external software which
	    annotates the map.
	
	--  Automatic declutturing of various objects in the map
	    by the Geographic Display Agent and automatic modification
	    of specified latitude/longitude limits to suit window
	    geometry.

	So far our solution is to encapsulate all map window manipulation
	within the GDA.  I'll be interested to see how PEX solves
	the various problems of input and X event multiplexing, and
	of sharing sufficient knowledge of the image it generates with
	an external software component.

---------------------
Paul Raveling
Raveling@vaxa.isi.edu