Path: utzoo!attcan!uunet!lll-winken!csd4.milw.wisc.edu!bionet!agate!ucbvax!tut.cis.ohio-state.edu!bloom-beacon!EXPO.LCS.MIT.EDU!rws
From: rws@EXPO.LCS.MIT.EDU (Bob Scheifler)
Newsgroups: comp.windows.x
Subject: Re: fix2 patch to window.c destroys backing_store
Message-ID: <8904132030.AA00612@EXPIRE.LCS.MIT.EDU>
Date: 13 Apr 89 20:30:31 GMT
References: <HIEBELER.89Apr13142610@turing.cs.rpi.edu>
Sender: daemon@bloom-beacon.MIT.EDU
Organization: The Internet
Lines: 28


      Now, I have tested my program which uses backing_store on a
    Sun-3/50, using a version of the X11R3 server with those 2
    lines, and without them.  With them, backing_store does NOT
    work.  Without them, it DOES work fine.

I suppose stranger things are true, but I find this hard to believe.

     -  Under what circumstances does this core dump occurr? 

When calling UnmapSubwindows on a window with no children.

    I.e. is it "dangerous" to run the server without that patch?

Not unless you have something that calls it.

    -  Is this patch really supposed to disable backing_store?

No.

    those 2 lines seem
      to say "if you're not the first child, don't do anything",

No, they say "if the window doesn't have any children, don't do anything".

    This seems to be hitting that return.

Have you actually witnessed this with a debugger?