PDA as "X terminal"

On Fri, 15 Oct 2010 22:44:52 +0100, Nobody wrote:

> There's already a VNC-like package which is implemented as a compositing
> manager on the application side, so that each window gets its own native
> window on the terminal side rather than one big window for the desktop.
> But I have forgotten what it's called.

"Xpra"

http://code.google.com/p/partiwm/wiki/xpra

Hi Theo,

Theo Markettos wrote:
> D Yuniskis <not.going.to.be@seen.com> wrote:
>> Can I configure it and the PDA so that it *locks* into this
>> mode?  I.e., nothing short of reflashing the device (or
>> something with a similarly high bar) can turn the browser
>> *off* (and, presumably, the browser *never* crashes)
> 
> Depends on what you run on the PDA of course, but I don't see why not.
> Assuming you can put Linux on the PDA, you just run NetSurf (or whatever) in
> your init script.  Then disable any keypresses to get into boot menus etc.

<frown>  I was trying to avoid reflashing the PDAs.
Of course, if I consider everything as "fair game",
then I can make the PDA do *exactly* what I want/need.

> "Browser" and "never crashes"?  Hmm, don't think I've ever seen one of those
> ;-)

Yeah, Doug's initial comment re: X being a heavyweight process
pales when you look at all the cruft in a typical browser!
E.g., In 25 years, I've never seen an X terminal crash; OTOH,
I can do that pretty easily with a browser in 25 *minutes*  :-/

>> JavaScript would only be necessary if I couldn't have some
>> other processes running alongside the browser (this would
>> be a kludge).  I am more than happy just pushing "drawing
>> primitives" at the handheld and waiting for input events.
> 
> That sounds like VNC... you don't have to use any particular windowing
> system over VNC, there are libraries to push drawing primitives through it. 
> So you just say 'rectangle here', 'rectangle there', getting back 'mouse
> click here' etc, and it's up to you to render pixels in the rectangles, deal
> with UI events etc.

The X core protocol does this.  Extensions to it help reduce
network traffic at the expense of client side caching, etc.

I think a *smart* use of the protocol would let me preload
resources into the client -- since I *know* a priori what the
"client's" capabilities will be -- and eleminate a good deal
of traffic without having to write any code on the "client" side
(e.g., send all the pixmaps over en masse and store them
on the X server; then call them up explicitly from the
application).

This has the advantage of getting me out of the "display server"
business entirely!