Simple responses:
1) network transparency is worth alot, particularly with network
attached PDA's. It means you can build a single application,
rather than one with both host and PDA side, and use
screens, mice and keyboards on your regular machines
when you need to.
BTW, network transparency is NOT where most of the code
space goes.
2) bug for bug compatibility for programmers is worth even more.
No retraining required.
3) We believe in Moore's law: while the iPAQ has relatively alot
of memory today, this will become the norm, rather than
the exceptional case.
4) there is more dieting to be done on the core X libraries.
Note that there is a different question as to whether the current GUI
libraries (e.g. gnome, etc.) have small enough footprints. I believe
Moore's law is most likely to help there, though I've seen data that makes
me believe some of those libraries (e.g. Gnome) can be greatly shrunk.
But smaller toolkit libraries may be needed: only time will tell.
I personally believe 16 meg of RAM isn't going to be an issue given our
current footprint. We'll be booting a H3150 shortly, and I expect things
will "just work", given the amount of free memory we currently have. 8
meg of ram and 2 of flash gets to be more of a challenge: dunno how things
will work out there (yet).
The IBM watch folks are using X with old style X servers and libraries,
and manage, somehow, to fit in 8 megabytes of RAM and 4 meg of flash.
I'm giving them a hand right now on how to build the new X server, and
we'll work jointly on shrinking the X libraries, which indeed are larger
than they should be.
- Jim
> Sender: ipaq-admin@handhelds.org
> From: Nils Faerber <nils@kernelconcepts.de>
> Date: Mon, 23 Oct 2000 17:22:17 +0200
> To: jg@pa.dec.com (Jim Gettys), xinchong <xinchongmbox@hotmail.com>
> Cc: ipaq@handhelds.org
> Subject: Re: [iPAQ] FLTK, NanoX and Microwindows?
> -----
> On Mon, 23 Oct 2000, Jim Gettys wrote:
> > FLTK is available on the iPAQ.
> Don't know about the X11 version but the nano-X version compiles alomost out
> of
> the box using the skiff cross utils.
>
> > I believe the Microwindows folks may have done a port.
> Nope, as it compiles cleanly. I don't know of any ready to go binary
> distribution. Only thing there is is an older version of viewml for ARM which
> supposedly has the nano-X and ftlk stuff statically linked in.
> Wasn't there some public diskspace planned for handhelds.org so that people
> can
> upload stuff for public download?
> If yes I would contribute iPaq ready FLNX and microwindows (nano-X) binaries!
We'd be happy to host it. Just create a SourceForge project and go
to it!
>
> > Much of the reason for Microwindows don't exist on the iPAQ, both
> > due to relatively generous memory size available on the iPAQ and the fact
> > that the X server we use is much smaller than "conventional" X servers,
> > being around 700KB right now (including Keith Packard's new rendering
> > extension, which provides antialiased text and graphics).
> > This makes it really easy to get applications up and running on the iPAQ.
> Yes, you are of course right.
> But is this really the way to go for Linux on PDAs?
> I would prefer a Linux PDA approach that alomost any PDA could use and X11 is
> for sure not suited for, for example, the Helio. With just 8MB RAM and 2MB
> Flash
> it will not fit. Nano-X would.
> You and many others put a lot of work into the X11 version which only a very
> limited number of PDAs can make use of. This will result in at least two Linux
> PDA trees, one for small scale PDAs and one for the bigger ones. Applications
> for typical PDA use are rare enough for Linux, why force this work to be done
> twice?
> Nano-X is at least one magnitude smaller than X11 when you take all needed
> libraries into account. With FLTK (or better FLNX) we have a working toolkit
> for it and can start right away.
> Nearly none existing X11 application can be used 1:1 on the iPaq or any other
> Linux PDA. You will have to tailor them to the new screen dimensions (much
> smaller than nowadays screens) or even rewrite them to some extent. So the
> easy
> porting argument is not fully valid. The network transparancy is IMHO not that
> important for a PDA. So what reasons are there left for using X11?
Different people have different opinions. I don't think doubling the
programming problem the way WinCE and PalmOS does is a good solution.
> OK, Nano-X is not really mature right now but has gained pace during the last
> about half a year. Assuming the same development speed I suppose that it will
> be X11 almost equivalent by the beginning of next year.
Good luck with it..
>
> Just my $0.02 ;)
> nils faerber
-- Jim Gettys Technology and Corporate Development Compaq Computer Corporation jg@pa.dec.comReceived on Mon Oct 23 08:53:01 2000
This archive was generated by hypermail 2.1.8 : Tue May 04 2004 - 09:43:44 EDT