[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 3 button mouse (was: Re: [MiNT] X magic)

> > > > I agree. To me an X server is a piece of software that allows you run
> > > > remote X window programs as if you were on the console, like telnet
> > > And even though XaAES is not an X server, it IS intended to be able to do
> > > just that. :)
> > 
> > AES-applications over network? I don't know about that, you'll need a
> > completely new VDI for that (perhaps a driver for fVDI?). And remember

A network driver for fVDI should not be hard to write. Naturally, you would
also need a 'server' on the other computer that would do the actual drawing.
Note that the drawing would not necessarily have to be done by an actual VDI
- it could just as well be a Windows or X program...

> > that unlike X, AES-applications are responsible for all redraws. X
> > knows the contents of it's windows, so it can do reasonable fast

X _can_ buffer windows, but AFAIK that's not mandatory. I don't know what
programs make use of it.

> > redraws, but constant redraws over a network (like AES/VDI would have
> > to do) would be pretty slow.

Not necessarily, but it would depend on what the program was doing.
Sure, the actual transfer would take some time, but on the other hand the
drawing would be totally decoupled from the application and could be done
at the same time as more drawing commands are sent.

Also, most redraws probably wouldn't need to send much data and you could
use various ways to cut down on the amount of it (compression or checks
against previously sent data for example).

> This is somewhat where the oAESis in the OSIS project is going. oAESis
> in Linux currently consists of a server and a library, which acts as a
> client to the server. All programs compiled with the library will send
> messages to the server to do things.

Originally, XaAES always operated in a kind of client/server mode, where all
TRAPs were converted into messages sent over a pipe (IIRC, only the actual
function number and an address to the data was sent).
Unfortunately, there doesn't seem to be a fast way to send messages under
MiNT (I'll have to check if perhaps the kernel entry modifications recently
mentioned has improved that), so this was often quite slow.
In later versions, the message passing could be turned off in favour of
direct calls to improve the speed.

> This brings some problems to oVDIsis, of course, because it has to
> share the workstations between processes. We're (CG) currently working
> on this, using shared memory. oVDIsis is only a library which oAESis
> and the programs are using.

Is there a Web page somewhere where one can find out what's happening
with oVDIsis?

fVDI currently doesn't worry about memory protection or even multiple
processes (normal VDI calls are supposed to be atomic), but it also is
supposed to be used as a (Fenix) library later on (so there are no
problems with reentrancy).

  Chalmers University   | Why are these |  e-mail:   rand@cd.chalmers.se
     of Technology      |  .signatures  |            johan@rand.thn.htu.se
                        | so hard to do |  WWW/ftp:  rand.thn.htu.se
   Gothenburg, Sweden   |     well?     |            (MGIFv5, QLem, BAD MOOD)