Path: utzoo!attcan!uunet!garfield!john13
From: john13@garfield.MUN.EDU (John Russell)
Newsgroups: comp.sys.amiga.tech
Subject: CloseScreen internals
Keywords: closescreen 1.4
Message-ID: <4996@garfield.MUN.EDU>
Date: 12 Nov 88 21:22:14 GMT
Organization: CS Dept., Memorial U. of Newfoundland, St. John's
Lines: 31

I'm curious as to whether the inner workings of CloseScreen could be
changed for the better without breaking anything. I think they could.

Why not make the screen structure's presence dependent on a semaphore?

OpenScreen: initializes semaphore value to 0
OpenWindow: increments semaphore value of that screen unless WBenchWindow
	flag set (so Open/CloseWorkbench work)
CloseWindow: decrements semaphore value; when all windows are closed
	semaphore reads 0
CloseScreen: decrements semaphore value; normally this is called after
	all windows are closed so value goes to -1

All functions that decremented the count would check, and if the count was
-1 would actually close the screen and remove it from the system.

The benefits? This would allow windows to open on any screen (maybe unbeknownst 
to the screen owner) and, if the owner attempted a CloseScreen it would succeed
but not actually take the screen away until the remaining window(s)
finally closed.

Detriments? I can't think of any; this is similar to the way it's supposed to
work now, except that it's the responsibility of the screen owner to close
all his windows before closing the screen. The system should provide just
that extra bit of security checking, thus opening up major opportunities for
shared screens between applications.

John
-- 
"Families of Murder Victims bring Human Face to Court Proceedings"
		-- Gets my vote as most poorly-worded headline of the year