Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!thunder.mcrcim.mcgill.edu!snorkelwacker.mit.edu!apple!usc!wuarchive!uwm.edu!linac!att!cbfsb!cbnewsc!gregg
From: gregg@cbnewsc.att.com (gregg.g.wonderly)
Newsgroups: comp.sys.amiga.tech
Subject: Re: Shared IDCMP ports
Message-ID: <1991Jan17.215456.4159@cbnewsc.att.com>
Date: 17 Jan 91 21:54:56 GMT
References: <1991Jan13.084228.25939@macc.wisc.edu>
Organization: AT&T Bell Laboratories
Lines: 35

From article <1991Jan13.084228.25939@macc.wisc.edu>, by gilmore@macc.wisc.edu (Neil Gilmore):
> I also had a problem with shared IDCMP ports, even though I was 
> (apparently) doing everything correctly.

> My problem turned out 
> to be that I was using the menu to select when to close a window, the 
> window which generated the message. Unfortunatly, I was using the real 
> message, not a copy, and replying at the end of an event loop.

> Now, I copy the 
> message and Reply() immediately, and haven't crashed since on closing a 
> window.

I had a problem like this too...  Now I realize why all of the RKM examples
are written as

	class = imsg->Class;
	gadgptr = (struct Gadget *)imsg->IAddress;
	...etc

	ReplyMsg( (struct Message *)imsg );

	switch( class )
	{
		case MENUPICK: ...  break;
		case GADGETUP: ...	break;
		...etc
	}

They really seem to be suggesting that you should reply to the packet before
you proceed!

-- 
-----
gregg.g.wonderly@att.com   (AT&T bell laboratories)