VNC optimized to almost 30% bandwidth raw (40% zlib)

Seth Cohn seth "at"
Mon, 28 Aug 2000 19:23:50 +0000

Wez wrote:

> > The function GetChangedRegoin (vncbuffer) looks for a changed line in a 
> rect
> > and then copy the full rectangle to the update buffer.
> > I changed the code so that only the changed rectangle is passed for update
> > and not the full.
 > ++  Is there a reason to use 32x32 rectangles and not a variable (8X32,
> > 17x32 ..)
> >
> > ++  I tested the code using zlib in debug mode on a single application
> > (internet explorer
> > showing a running text in java)
>There is a reason for doing this - it avoids the case where a large number
>of small regions have changed all over the screen, which would result in a
>huge number of update rectangles, swamping the network connection.
>It's a feature that assumes most rectangles will not be RAW, so that
>reducing the number of rectangles in an update will reduce the bandwidth by
>reducing the overhead.

Hmm... If he's seeing _that_ much of a bandwidth drop, maybe this should 
become an
_option_, similar to choosing some of the other display option?  For 
displays with lots of small region changes, the default setting, but for a 
speed optimized display when the changes are minor... this patched 
behavior.   am I way off here?

To unsubscribe, send a message with the line: unsubscribe vnc-list
to majordomo "at"
See also: