On Sat, 12 Jan 2002, Ned Konz wrote:

> Hi all,
> There was some discussion in '99/'00 about having Squeak natively speak VNC.
> I just tried out Squeak talking to a Linux Xvnc server, with a client on a 
> slow handheld over the network.
> 
> I found that there is about a 1.5 to 2 second delay between typing a
> character and seeing it on the display, which is obviously
> unacceptable for normal uses. I don't know what I could do to speed
> this up, or to reduce memory. So having Squeak speak VNC directly is
> promising.

I ran it over an ISDN line successfully (using an 8 bit Xvnc server) - are
you sure the delay could be reduced by a native RFB implementation?

> I know that Adam Feuer and Stephen Pair both did some work on this, as well 
> as Bert Freudenberg.
> 
> Is there any interest in doing this collaboratively?

Hmm, I've got no spare time for this now ...

> I looked at the Xvnc code, and there's way too much there. My feeling was 
> that it would be easier and cleaner to start from the protocol documentation 
> and do a clean port, using the Xvnc code as a reference only when and if 
> necessary.

I think a clean implementation of the RFB protocol shouldn't be too hard.  
We didn't really get to a conclusion of where to place the hook, though -
ioShowDisplay() in the VM would be the most general, next was hooking
BitBlt's copyBits, the last, for Morphic only, writing an
RFBRemoteCanvas...

-- Bert