Replied: Wed, 09 Nov 1994 10:15:01 -0800
Replied: Olof Samuelsson <d8samol@dtek.chalmers.se>
Received: from thuk.dtek.chalmers.se by paris.ics.uci.edu id aa02183;
          8 Nov 94 18:25 PST
Received: (from d8samol@localhost) by thuk.dtek.chalmers.se (8.6.9/8.6.9) id DAA03320; Wed, 9 Nov 1994 03:25:24 +0100
Date: Wed, 9 Nov 1994 03:25:24 +0100
From: Olof Samuelsson <d8samol@dtek.chalmers.se>
Message-Id: <199411090225.DAA03320@thuk.dtek.chalmers.se>
To: hadley@dv.ICS.UCI.EDU
Subject: Re: BRMH2.0p14.solaris, buggy on mono X11R6 
In-Reply-To: <10767.784341156@dv.ics.uci.edu>
References: <199411081113.MAA17592@varoom.dtek.chalmers.se>
	<10767.784341156@dv.ics.uci.edu>

Tedd Hadley writes:
> In message <199411081113.MAA17592@varoom.dtek.chalmers.se>, Olof Samuelsson wri
> tes:
>  > When using BRMH2.0p14.solaris on a mono screen, the tractor lines are
>  > not erased properly when the T/P is used (i.e. it leaves garbage on
>  > local), when continueTractor and showTractorPressor is on.
> 
>    This problem has me at wit's end.  At one point here, all the
>    clients started leaving garbage on the screen from T/P.

All the clients?  For all architectures?  (You seem to have a lot of
them at uci, btw)  My first guess, an incompatibility between R6 and
older X11s on Solaris is dead wrong then.

>    I wrote a quick test program to highlight the problem (I'll
>    include it for you) and discoverd that erasing dotted lines with
>    solid lines left random pixels here and there.  So I changed the
>    client to explicitly erase dotted lines _with_ dotted lines.  How
>    can you go wrong with that I thought, but apparently not.  Let me
>    know what the test shows.

I compiled the test with both the X11R6 and the X11 OpenWindows libraries
(those should be R5, I think .. we use no OpenWiindows stuff here),
and .. tried it on both a color and a BW screen, and in none of the
combinations managed to get any traces of dotted lines left in the
window.  Nothing.  I saw them being drawn (I linked with libucb, I
don't trust anything linked with libucb..), but no left-over pixels at
all.


>    Right, this just clears the entire window, a little inefficient for
>    slow displays and might "blink" a little.

As an interrim solution, put fastClear on by default for mono screens?

>  > Has anyone asked for a statically linked Solaris2 version?  Now I'm
>  > handediting the binary to change the library references to the X11r6
>  > libraries instead of the r4(?) the client wants.  The client works
>  > fine, but I'm just waiting for it break :-)
>    
>    I can build a static binary of that (maybe the test program will
>    give some insight and we can try a few drawing type permutations) or you 
>    can set up a shell script or alias of the form
> 
>    (setenv LD_LIBRARY_PATH /to/r6/lib ; exec netrek $*)

Didn't think of the OpenWindows libraries ... I'll get a fresh binary
and check whether Solaris can take libX11.so.4 instead of
libX11.so.4.22, which is the library reference in the client, I think
.. tomorrow.

>    Let me try 
> 
> 
>    (do 'xmkmf ; make' to compile.  Usage:

Not that easy :-)  Had to tell it to link with libucb, socket and nsl
too, this is Solaris after all :)



I'd like to compile the client here, have a brief look at x11window.c,
and see what happens in the client.  However, that will have to wait,
it's very late here, and I have (another) busy day tomorrow.  Hmm,
maybe I should merge in that silly playerlist I talked about too
... it's a few months late now...

Maybe you should try compiling xtest, and mail the binary to me, it
would be the most similar test setup we could do.

I hope the results helped you somehow, I just got more confused :-)

Have nice afternoon,
/Olof
--
