Path: utzoo!attcan!uunet!cs.utexas.edu!usc!snorkelwacker!bloom-beacon!EXPO.LCS.MIT.EDU!rws
From: rws@EXPO.LCS.MIT.EDU (Bob Scheifler)
Newsgroups: comp.windows.x
Subject: Re: Shape extension is slooooooooooooooow
Message-ID: <9002221357.AA01043@expire.lcs.mit.edu>
Date: 22 Feb 90 13:57:32 GMT
References: <9002220532.AA22951@Larry.McRCIM.McGill.EDU>
Sender: daemon@athena.mit.edu (Mr Background)
Organization: The Internet
Lines: 27


    On a Sun-3 with the MIT sample
    server, I find shape to be very slow as well.

Under "normal" use I find it to be quite acceptable, at least on machines
with decent frame buffers and decent CPUs (a Sun 3/60 does qualify).

    I have a terminal
    emulator which I hacked on to make the window transparent (actually,
    nonexistent) where characters aren't.

This is not normal use.

    Scrolling is agonizing.

I'm not surprised.

    To be more precise, *changing* shapes is very slow.

Yes, especially for complex shapes, in the current implementation.

    Is this inherent in the extension or what?

Probably not, it's probably largely due to the region representation used
in the server.  I'm sure much more complicated and clever representations
are possible.  Perhaps you could perform the same experiment with something
like psterm in NeWS, and see if it has similar problems?