Path: utzoo!utgpu!water!watmath!clyde!att!rutgers!gatech!bloom-beacon!think!ames!oliveb!amiga!jimm
From: jimm@amiga.UUCP (Jim Mackraz)
Newsgroups: comp.sys.amiga
Subject: Re: Intuition Bug.
Message-ID: <2784@amiga.UUCP>
Date: 23 Aug 88 19:14:30 GMT
References: <8808200650.AA14865@jade.berkeley.edu> <65150@sun.uucp>
Reply-To: jimm@cloyd.UUCP (Jim Mackraz)
Organization: Commodore-Amiga Inc, Los Gatos CA
Lines: 78

In article <65150@sun.uucp> cmcmanis@sun.UUCP (Chuck McManis) writes:
)In article <8808200650.AA14865@jade.berkeley.edu> (Jonathan Crone) writes:
)>Hi all, could someone Email me the exact situation that
)>causes that problem that Intuition has with interlaced and
)>non interlaced screens???
)>I'm trying to isolate a bug in something and I'm not sure what
)>it is or whether its a problem in Intuition....
)
)It isn't Intuition :-). Actually, the only bug that I know of has to do
)with ReMakeDisplay() when operating on interlace screens that aren't the
)front most screen. To see this in action open two interlace and one non
)interlace screen and then use the screen to {front|back} gadgets in the
)corners to rearrange them. Eventually you will see a screen that has 
)obviously got a broken copper list. Do you call ReMakeDisplay() in your
)code? 
)
)--Chuck McManis

Well, it is Intuition, Remake/RethinkDisplay().  What happens in
RethinkDisplay() is that only those copper lists that need changing
are to be rebuilt.  When dragging a screen, this is at most two screens:
the one getting taller and the one getting shorter (the two on either side
of the "seam" being dragged").

Now what happens is this:
RemakeDisplay() creates a single new copper list for an interlaced viewport
if that viewport is hidden and the whole display is non-interlaced.
When you pop that screen to the front, RethinkDisplay() is called.
The Viewport height hasn't changed, but everyone's copper lists are remade
when the screen switches into interlaced mode.

The problem happens when in the intervening time, some other interlaced
screen appears, the display goes into interlaced mode, and our
interlaced viewport's non-interlaced copper list still hasn't
been rebuilt. 

Now when it pops to the front, it passes the height change check,
it passes the interlace-change check, so it isn't remade.  The bug is
that VP_HIDE changing hasn't caused a further incentive to rebuild
the copper list for our interlaced screens viewport.

Then some magic happens when you use copper lists created for the wrong
purpose.  Note that a key point is that the mode of the View a viewport
is hanging off of at the time of MakeVPort() determines how the
copper lists will be built, in particular, two (for interlace) or one.


So although the bug is in RethinkDisplay(), the problem arises when
you call RemakeDisplay().  So don't do that if you can help it.

A hack: try jamming a zero into the DHeight field of each screens
viewport which is showing VP_HIDE in its modes, after you RemakeDisplay().
That way, when visible, RethinkDisplay() will rebuild its coppuppies.

This hack is by no means guaranteed to work in V1.4, but I'll try.
For that matter, it isn't guaranteed to work in V1.2/V1.3.  It's just
a thought.

Better yet is not to call RemakeDisplay() if you don't have to, and
you usually don't.  RemakeDisplay() is only used to construct new copper
lists for every viewport.  You rarely need to do that.

The standard sequence when you want to screw over your screen's viewport
is:
	MakeScreen( screen );	/* construct new coppups for my vp	*/
	RethinkDisplay();	/* mrgcop and loadview			*/


Do this, over the hack, if at all possible.

	jimm


-- 
	Jim Mackraz, I and I Computing	  
	amiga!jimm	BIX:jmackraz
Opinions are my own.  Comments regarding the Amiga operating system, and
all others, are not to be taken as Commodore official policy.