
|
[haiku-appserver]
||
[Date Prev]
[10-2004 Date Index]
[Date Next]
||
[Thread Prev]
[10-2004 Thread Index]
[Thread Next]
[haiku-appserver] Re: Backbuffer access
- From: "Mikael Jansson (mailing lists)" <lists@xxxxxxxxxxxxxxxxx>
- To: haiku-appserver@xxxxxxxxxxxxx
- Date: Sat, 23 Oct 2004 14:42:09 GMT
"DarkWyrm" <bpmagic@xxxxxxxxxxxxxxx>:
>
> > Okay. I was just thinking (not really knowing how the app_server
> > is
> > layered) that just providing the developer with a shared area or so
> > to
> > use. Main point, I guess, is that /if/ someone wants to provide for
> > such a structure for the R1 app_server, it doesn't need to be
> > completely rebuilt, just to hook it in in the approriate layer and
> > you're set, basically. But, like I said - I don't know the layers
> > very
> > well, so it's hard for me to tell.
> It wouldn't be all that hard to hack the server, really. There are a
> few methods I was kicking around but they all basically centered
> around
> hooking into a certain part of the server to notify a VNC server of a
> screen update which could then notify any clients attached. There are
> a
> bunch of different ways that it could be done -- tweaking
> DisplayDriver
> or playing around with the DispatchMessage functions in ServerApp and
> ServerWindow are a couple I've toyed with. They could be used to do
> the
> notifying of the VNC server which could then notify the client(s).
>
Okay, that's nice. As for the actual protocol, I'm not sure how they
differ, but IIRC, RDP uses "glyphs" to achieve a higher throughoutput.
Would it be possible to create a proxy that sent the app_server
commands _themselves_ over the network to another computer sitting
there with its own app_server and then doing the drawing? For bitmap
data it'll be the same as a VNC-style program, but for drawing strings,
stroking lines, fills and such, it would be a _lot_ faster. What would
be neded, of course, is to have app_server running in a window on a
BeOS/X/Win32 system. I'm not sure how flexible app_server is in this
regard, but would it be possible to write an accellerant that drew on
an SDL surface? Because in that case, we'd have a highly viable cross-
platform solution.
I hope I'm not annoying you guys too much. I appreciate the work you
do, and I don't understand all of it, so my suggestions might be a bit
off.
--
Mikael Jansson
http://mikael.jansson.be
|

|