Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!bloom-beacon!ZERMATT.LCS.MIT.EDU!RWS
From: RWS@ZERMATT.LCS.MIT.EDU (Robert Scheifler)
Newsgroups: comp.windows.x
Subject: Bug in reported events from X11
Message-ID: <19881028210626.5.RWS@KILLINGTON.LCS.MIT.EDU>
Date: 28 Oct 88 21:06:00 GMT
References: <881026-180534-14387@Xerox>
Sender: daemon@bloom-beacon.MIT.EDU
Organization: The Internet
Lines: 28


    Date: Wed, 26 Oct 88 18:05:06 PDT
    From: jacobi.pa@xerox.com

    I move the cursor, from background into "child", push a button, move 
    the cursor into "other" and release the button.

    1) "child" has 2 leaveNotify events.  

Correct.  But, you need to look at the "mode" component of the events.
One of them is Normal, one of them is Ungrab.  The client can
distinguish if it cares.

    2) "parent" did not get the leaveNotify before the button was released. 
    3) "other" did not get the enterNotify before the button was released. 
    4) "other" did not get a buttonRelease event at all. 

Correct, because the pointer has been grabbed by "child", and the grab
affects the event distribution.

    1 looks like a bug to me.

No.

    2-4 could be declared a feature.  However, I wouldn't like that.

It is considered a feature that an automatic grab occurs when the
pointer is pressed.