RE: [Java-gnome-developer] Development Status

[Almotasim <alm...@no...>]

Tks for your answers.=20
I'm happy to read that there's a lot of synergy between Java-Gnome and
SWT as I was fearing that too many similar-but-not-really-compatible
projects were being developped.=20

I have several remarks and questions divided below in remarks regarding
strategy and remarks concering technical points.=20

1) Strategy=20

Is it so much useful to maintain Windows+Motif support ? Gnome on the
desktop is making huge progress, whereas Windows doesn't innovate much
and is therefore liable, in my opinion, to disappear in the mid-term (5
years?), because the system itself is conceptually too poor to compete
with the various OS that the open-source community + companies like IBM
or Sun have coded in the last 20 years.=20

2) Conception/technical points=20

>A higher-level object-oriented API is needed in order to make
>java-gnome really viable.=20
I think that's a very good idea to build a high-level OO GUI-API
gathering the best features of Swing, Gtk, Qt, MFC + more (I actually
thought at first it was already part of the Eclipse project). The key
point probably relies on the community such a project is able to gather;
I believe many engineering schools (such as http://www.emn.fr/ in
France, which works closely with OTI) would be interested in joining the
project both at a research level and at a coding effort level; hopefully
OTI team would also participate strongly, making the "high-level API"
more quickly available for commercial-apps use. I guess the industry
will need more and more advanced widgets, such as hyperbolic trees
(http://treebolic.sortilege.net), smart tables like those in Gnumeric,
advanced visualition tools etc., hence the availability of such an API
could help the industry working better.=20

Which language in the mid-term for OO programming: Java, Python, C++ ?
Is there any reason to keep using so many very close languages instead
of focusing on theorical problems related to OO programming?
(meta-objects protocols, automatic components assembling etc.). I'd be
happy to know what Ximian developpers think about the OO language
question, as they're making very good apps based on C/Gtk and Bonobo.
Java is a cool language but I'm wondering sometimes if these guys are
not going to produce a new LGPL Java-like language. I believe Java
should now be released in true open-source, so that designers and
developers from all companies could focus on a cooperative effort,
without developing so many language-gateways (of course diversity is
very important but here it looks like there is no real conceptual
difference between Java, C++, Python, etc., or is there?).=20

One last remark: Gnu/Linux is still missing the game in the games
software area; it's obvious that encouraging the development of games on
Linux could help creating collectively advanced high-level generic
OO-APIs for open-source graphical environnements ("intelligent" games as
far as possible of course).=20

Regards=20

Almo=20





Le lun 15/07/2002 =E0 16:54, Jeffrey Morgan a =E9crit :=20
> > Is there a possible synergy between Java/Gnome and IBM-SWT library
> > (eclipse.org)? What's Java-Gnome position (technically and
> > strategically) regarding IBM-SWT/JFace, which seems to run well on
> > Gnome2.0 ?=20
>=20
> There is a lot of synergy between these projects.  In fact, I am
> currently working on the SWT project at the same time as working
> on java-gnome.  There are a few problems with SWT that are being=20
> addressed.  They are:
>=20
> * SWT uses the GTK 1.x tree and list widgets (SWT team is working on this=
)
> * SWT uses the GTK 1.x text widget (I am working on this)
> * SWT/GTK does not have a consistent approach for NLS (SWT team is workin=
g
>   on this)
> * SWT relies on being able to utilize z-order.  GTK doesn't support z-ord=
er.
>   The fix that was put in place is probably contributing to poor
> performance.
>   (This is an issue that I intend to bring up at this week's GNOME Summit=
)
> * The SWT and GTK model for Composite/Child (SWT) and Container/Child (GT=
K)
>   are very different.  As a result, SWT/GTK was forced to introduce hacks
>   to handle widget sizing and positioning.  This is contributing to the p=
oor
>   performance.  (I also hope to bring this up at the GNOME Summit)
> * SWT doesn't have close to full coverage of the APIs that are provided b=
y
>   GTK.  SWT also doesn't provide any coverage for the GNOME APIs. (See my
>   proposed solution below)
>=20
> I intended to wait until my next release to put a proposal out to the
> community but since you have touch on this topic I think I will do it now=
.
> SWT will continue to mature at a rate that is much faster than java-gnome
> due to the fact that IBM/OTI has full-time resources working on it.
> Also, due to the fact that SWT runs natively on several platforms it will
> gather much greater industry support.  At a low level, the way SWT is
> accessing
> the native peers is superior to the approach taken by java-gnome.  It wou=
ld=20
> appear that SWT would be an ideal choice for somebody wanting to write a=20
> java application that will run on GTK.  You would also have the added
> ability=20
> to support Motif, Win32, OSX Carbon, and other platforms.
>=20
> The primary problem with SWT is that it is not currently intended to be
> a complete/robust GUI framework.  Not all widgets supported on the host
> operating systems are supported on SWT.  Also, although JFace does provid=
e
> some higher-level facilities it is primary focused on providing a framewo=
rk
> for the eclipse platform.  At the same time, java-gnome attempts to map
> directly to the underlying APIs provided by GTK and GNOME.  This makes
> the java-gnome class library huge and very complex to use. A higher-level=
=20
> object-oriented API is needed in order to make java-gnome really viable.
>=20
> My proposal is for a new project (or a significant morph of java-gnome).
> This new project would use SWT as it's foundation and would address the=20
> above mentioned issues.  This new project would not only focus on GTK. =20
> It would also provide support for Motif and Win32 (possibly others). =20
> Perhaps over time some of the code produced by this new project could=20
> be incorporated into SWT and some of it would stay independent of SWT. =20
>=20
> A proposed task list to start the project follows:
>=20
> 1) Identify lead developer for each of the platforms:
>    GTK - Jeff Morgan
>    Motif - ??
>    Win32 - ??
>=20
> 2) Define the new API and create an object model to be presented to the
>    community:
>    -I have spent a lot of time this year studying various GUI frameworks
> (QT,=20
>     GTK, Motif, MFC).  I would like to begin a discussion with the
> developers=20
>     on this new project to decide what functions should be included in th=
is
>     new GUI framework.  Perhaps it would be a combination of the best
> features=20
>     of several of the mentioned frameworks.  The result of this discussio=
n=20
>     would be an object model and a plan for development.  These artifacts
>     would then be presented to the community at large for comments and=20
>     suggestions.
> 3) Once the design and plan are finalized create begin the effort.
>=20
> > Do you know what Sun strategy is regarding Java integration into Gnome
> > (Solaris will use Gnome as Desktop Environnement, but then which
> > language will Sun recommend to build GUIs - as far as Swing=20
> > is obviously
> > not fast enough to build advanced smooth GUIs ?
>=20
> Sun would have to answer this question.  At one time there was
> much discussion about providing GTK peers for AWT.  I don't know
> where this discussion went.
>=20
>=20
> I would be very interested in hearing from the users of java-gnome
> about the proposal put forth above.  Although it is a significant
> departure from the current direction of this project I would be
> interested in your views if it is the correct direction.
>=20
> -Jeff