linuxcaffe; distros and desktops
Taavi Burns
jaaaarel-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org
Fri Oct 29 19:01:39 UTC 2004
On Fri, 29 Oct 2004 00:02:12 -0400 (EDT), Robert Brockway
<rbrockway-wgAaPJgzrDxH4x6Dk/4f9A at public.gmane.org> wrote:
> Hi Taavi. What do you mean by True transparency composited on the client?
AFAIK all of the currently implemented transparency schemes involve heavy
communication between the X server and the X client: the X client does the
rendering, fetching current screen state from the X server, and then dumps
the composited image back to the X server. I wouldn't want to do that on
the wire.
When things get to the point where the X client program just says "This is my
alpha channel" and the X server composites everything, then the speed will
be quite acceptable. Using proper compositing like this has other advantages,
too (like the elimination of 'tearing' when dragging one window across another,
because no application ever has to redraw itself because as far as
it's concerned,
it's never obscured.
Things were done all ugly-like in the past because things were rendered direclty
onto video memory, which was all being displayed on the screen. I
seem to recall
hearing that some of the high end UNIX workstation video cards got around this
with multiplane setups and stuff...but this also tended to limit
colour depth...I think.
I'm probably in way over my head at this point in the discussion.
> Do you mean the WM should run on the client? It can be done (NCD used to
> offer it as an option) but I don't see it making much of a difference in
> performance. I've been a user of thinclients for years and have been very
> happy with performance. Performance of thinclients is often underrated I
> find.
Nope, the WM can run on the terminal server. I was only speaking of
transparency
being computed on the thin client (X server) vs the
server+lotsofchatteronthewire
(X client).
--
taa
/*eof*/
--
The Toronto Linux Users Group. Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml
More information about the Legacy
mailing list