Sounds about right. I guess you'd send the params via socket, so you'd
use DXRegisterInputHandler to register your renderer on the socket fd;
then, whenever an update is received on the socket, your renderer would be
called. I'd put it in a loop internally checking for more input, and only
return from the handler when there's no more updates ready on the socket.
You might have to fiddle a bit to avoid falling behind the input stream -
when you go back to read the socket, throw away all but the most recent
sets of parameters. Do you render once per wall? I wonder about getting
the walls to swap buffers reasonably simultaneously.
I'll bundle up some code in my spare time and send it out.
Greg
Kent Eschenberg
<[EMAIL PROTECTED]> To:
[email protected]
Sent by: cc:
[EMAIL PROTECTED] Subject: Re: [opendx-users] DX
in "CAVE" Under W2K
son.ibm.com
05/30/2002 12:23 PM
Please respond to
opendx-users
Greg,
Thanks for your advice. I would indeed like a copy of the code you
mentioned.
In our setup, a separate process tracks the CAVE operator and generates a
list of view parameters, one per screen. I hope to leave this as is, and
merely change it so that the parameters are sent to DX.
In DX, it seems like the thing to do is to make a new version of the
Display module called DisplayCAVE. Inside here, there is probably one place
where all the objects are rendered. That area could be put in a loop with
one iteration per view. At the top of each loop, the DX camera would be
ingored in favor of the parameters send by the tracking process.
Of course, there are many other details. Does the above basic plan make
sense?
Regards,
Kent
- - -
Kent Eschenberg [EMAIL PROTECTED]
Scientific Visualization Specialist
Pittsburgh Supercomputing Center, CMU, Pittsburgh, PA
