discuss-gnustep
[Top][All Lists]
Advanced

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

Re: Embedded GNUstep - new rumours about embedded MacOS X and call for s


From: Richard Frith-Macdonald
Subject: Re: Embedded GNUstep - new rumours about embedded MacOS X and call for support
Date: Sat, 23 Dec 2006 14:40:30 +0000


On 20 Dec 2006, at 15:47, hns@computer.org wrote:

Hi all GNUsteppers,
there is new rumour about an embedded MacOS X (aka OSXlite, OSXmobile
etc.):

http://www.macrumors.com/pages/2006/12/20061219140034.shtml
http://www.appleinsider.com/article.php?id=2313
http://www.looprumors.com/LoopBlackPopup.php?mac-mobile,4046720790

All this shows where the industry is heading to: enhanced full
operating systems that run on (open) Smartphones and Settop boxes. And
it shows what the future needs: something besides Symbian, Qt and
Windows Mobile.

Therefore, I would like to take this as the opportunity to make a new
"call for support" for the embedded GNUstep fork "mySTEP". The goals of
mySTEP are exactly those described by the articles that I have cited
above: provide a stripped down version of GNUstep that is 100%
compatible to a Mac with Cocoa and is heavily optimized to run with the
memory and processor constraints of a Linux based PDA or Smartphone.

The status of mySTEP is already quite good: most Foundation and many
AppKit classes are working well on all Sharp Zaurus models and on Acer
n30 with Letux Linux  - and there are already two Linux based
Smartphones we want to port to next year and a third one is a good
candidate. Some recent developments are native/binary NIB loading,
@interface-compatibility of all classes to MacOS X 10.4, support for
shaped windows (e.g. rounded corners), a simplified WebKit, NSIndexSet,
NSIndexPath, basic PDF printing, and (not yet working) a complete
rewrite of the DO system to fully support NSMessagePort and
NSSocketPort, and too many others to list here.

Compared to GNUstep, QuantumSTEP is a stand-alone system including
application suite and window manager. It provides all that is required
for personal information management, mobile mail, web browsing, input
managers etc. on a handheld device. And sometimes it includes even the
X11 server (which is not readily available for a Zaurus). So it is more
like a mySTEP based "GPE".

Making mySTEP a smooth, reliable and finally useful system is something
different and therefore, if you are interested in supporting this
project, please start to look into the source code, submit patches,
contact me by mail - or during the planned presentation at FOSDEM 2007.

Might I suggest unforking ... ie merging your work into GNUstep with the aim
of eventually repositioning QuantumSTEP to use GNUstep.

It seems to me that the aims of the two projects are almost identical ...
both aim to be MacOS-X compatible, and while GNUstep is not
specifically targeted at handhelds, it *is* and always has been
targeted at cross-platform portability. A compiler/configure-time option to
optimise for handhelds would be a simple addition (if necessary).

My impression is that in some areas QuantumStep is more MacOS-X
compatible than GNUstep, and in other areas GNUstep is the more
compatible ... probabably because, despite your concentrated work on
improving MacOS-X compatibility, you started with mgSTEP as a base
and that is a very old fork which threw away a lot of MacOS-X compatibility code and a lot of performance optimisation code in the name of simplicity.
Thus you are forced to reintroduce missing code and to reoptimise for
performance on handhelds.

From a development point of view, it makes sense to work on GNUstep
and backport changes to mySTEP as necessary ... this is because the
GNUstep project is not just free/open-source software, but employs an
open project management philosophy too ... being hosted on a public
subversion source code control system, with clear coding standards
and with ChangeLog files so that it is simple to track changes.
For this reason. it is relatively easy to locate relevant changes in the
GNUstep code-base and then backport to mySTEP,  while working in the
opposite direction is much less easy.

I would be very happy to help with this if done in such a way as to benefit
GNUstep and reduce duplication of effort ... I certainly see running
GNUstep/QuantupSTEP on handhelds and generally for embedded
applications as a good thing.

One area I want to see addressed in GNUstep is theming .... not just for
the people who like eye-candy, but also for people who want to fit in
with the native look of applications on platforms like ms-windows, and
(relevent here) to fit well with different display sizes etc.






reply via email to

[Prev in Thread] Current Thread [Next in Thread]