Path: utzoo!mnetor!uunet!husc6!bloom-beacon!tut.cis.ohio-state.edu!mailrus!ames!pacbell!att-ih!alberta!myrias!cg
From: cg@myrias.UUCP (Chris Gray)
Newsgroups: comp.sys.amiga
Subject: Re: NOISYREQ anomaly
Message-ID: <571@myrias.UUCP>
Date: 16 Mar 88 19:48:49 GMT
References: <23669@amdahl.uts.amdahl.com> <24731@amdahl.uts.amdahl.com>
Reply-To: cg@myriasa.UUCP (Chris Gray)
Organization: Myrias Research Corporation
Lines: 17
Keywords: VT100 NOISYREQ window close requester
Summary: Requesters SHOULD block all other events

Tony Sumrall (acs@amdahl.uts.amdahl.com) suggests that he wants to be able
to see a window-close event when a requester in that window is active.
I so NO to this!! One of the points behind requesters is that they must be
handled before any other events for the window are. I've written a couple
of library routines (e.g. a palette editor) that can simply be passed a
pointer to the window, and will do their thing without any other knowledge
of what the window is doing. (I properly save and restore the IDCMP flags.)
When I use them, the other gadgets in the window do not even respond to
mouse actions, and this is precisely what I want - they are not currently
accessible, and not responding at all shows this to the user. If you want
out-of-band signals when you are asking for a file name, then use a second
window and merge the IDCMP streams from all of the ones you are using - don't
change the way requesters work!

-- 
Chris Gray		Myrias Research, Edmonton	+1 403 428 1616
	{uunet!mnetor,ubc-vision,watmath,vax135}!alberta!myrias!cg