Path: utzoo!attcan!uunet!wuarchive!gem.mps.ohio-state.edu!apple!bloom-beacon!EXPO.LCS.MIT.EDU!keith
From: keith@EXPO.LCS.MIT.EDU (Keith Packard)
Newsgroups: comp.windows.x
Subject: Re: transparent windows
Message-ID: <8909132016.AA20228@xenon.lcs.mit.edu>
Date: 13 Sep 89 20:16:16 GMT
References: <8909132031.AA02321@amd-26.hac.com>
Sender: daemon@bloom-beacon.MIT.EDU
Organization: The Internet
Lines: 24


> I am writing an application that requires "transparent" windows.
> I am hoping to achieve this by mapping multiple screens to the same
> display surface.

This is certainly not the model I would suggest for providing transparent
windows.  A better design is to create a new visual type (transparent
static gray or whatever you have), create a trivial extension which returns
the visualid of this new visual type, and the magic pixel value which
makes it transparent, and then create windows using the new visual type.
Automatically, you get cursor positions relative to both windows, you only
get one cursor and everything else works out fine (Enter and Leave
events work as expected).

Very small server changes will yield the desired results.  I suspect some
minor changes in ValidateTree will be required to get the correct
exposure events sent to the windows.

Please keep the mistakes of the past (the multiple screen hack for the
Sun 3/60 CG4 display) from clouding your judgement about how transparent
overlays should work.

Keith Packard
MIT X Consortium