scope-dev

[Scope-dev] Re: [jdistro-devel] Where do we go ?

[Christian H. <chr...@tu...>]

(Again a cross-posting ... but this is an important integration issue.
I think we all develop and develop but inter-communicate too little.)

Hi Gerard,

what was the reason for many of our projects to get started?
- JOS: an Operating System written in Java
- JDistro: a Java Desktop and Distribution containing several applications
- Scope: an extended MVC framework for Swing/JSP application creation
- CYBOP: a generic Java framework for system/application development
- ResMedicinae: a Java-based Medical Information System

Of course, there are many similar projects but what's common to the
projects above is:
- they are GNU GPLed (sorry, Scope is BSD)
- they use the Java programming language

My personal (lifetime) visions (dreams) are:
- a flexible operating system which can run on any device
- which allows in a very simple way to code input/ output modules
(textual/curses, graphical, web, vocal, Braille etc.)
- which already provides derived applications with basic communication
mechanisms, their protocols etc. (RMI, CORBA, SOAP, JDBC) so that an
application just sets a flag for which mechanism to use and that's all
- which integrates a nice desktop environment
- and all kinds of useful applications
- especially a complete Medical Information System including an EHR --
Electronic Health Record (the reason why I personally started free
software development)
- this all in an easy, clear, error-minimized way of programming
(which is why I had chosen Java as my favourite; possibly a new
language, for example based on XML, in the future -- Java is not "free"!
Python gains more and more popularity)

How do I think we all can contribute to that (or a similar) vision?
- continue to work on our projects
- start to look yet more into our neighbour projects
- try to find possible points of cooperation and communicate them

JOS:
- Did you think about how to allow a Graphical Desktop to base on JOS?

JDistro:
- Did you try to build on JOS or to integrate it into your distribution?

Scope:
- It seems to be very quiet in your mailing list. Did you all give up
or continue to contribute to the Free Software Community?
- Did you encounter design problems in Scope? If yes, then which?
What can we all learn from it?
- If the design turned out to be incomplete: What should be changed?
- If the project has stopped: Did you try to find a similar project to
bring in your knowledge and continue your work?

CYBOP (the questions I frequently have to ask myself):
- Did you loose focus, starting off from a small medical application
(ResMedicinae), ending up thinking about a generic framework for all
kinds of systems, about Operating System issues and even about a
replacement for the Java programming language?

All:
- Do you try to steadily improve your architectures and to be open
for new, effective ideas?
- Are you fixed on Java only or open to replace it wherever it turns
out not to be the ideal solution?

I can give you an answer to the CYBOP-questions, why I am where I am
now and how I came here, starting with one single line of Java code.
Can you answer the questions on your projects?

All I want is that we try to collaborate a bit more. Whenever something
seems unclear to me, I just ask the neighbour lists, for example some
months ago about Java event handling interna which were unclear to me.
Meanwhile, I could replace it in CYBOP by my own signal handling.
(I don't like the callback stuff with interfaces etc.; they cause too many
bidirectional interdependencies within a framework/ application;
but that's another topic.)

Concrete for JDistro:
- If the Desktop runs fine as it is, then just release your 1.0!
- Knoppix is a good idea; I think JOS was thinking about the same?
- Where do the console applications in JDistro and JOS differ, what
functionality is the same? Can this effort be merged?
- You mention a "JDistro library so that applications can be developed"
which sounds much like an "application framework" to me. Did you
inspect Scope and/or CYBOP to use their ideas or, better yet, to build
on these frameworks? Which of their ideas would be good for JDistro?
- The "virtual File System" is truely a duplication effort since JOS
as potential operating system has developed one, too.

Christian


On Wednesday 02 July 2003 10:31, Gerard COLLIN wrote:
>  Hi,
>
>  I like being part of this project, which I think is one of the best Java
> project available (and I'm using it everyday).
>
>  But I think something is missing: A goal.
>
>  What is the goal of JDistro ? Technically, it's to run all java
> applications in the same JVM under the same Desktop. That's fine, and I
> think most applications (except the heavy ones) now runs fine. But this
> could go one forever !
>
> In short, what needs to be in JDistro so that we can release a 1.0 ? What
> needs to be there ?
>
> I would like to discuss here what the result would be ? Should we make a
> JDistro CD using KNOPPIX ? If so, what progress is made in this area ?
>
> Is it a goal to make a JDistro library so that application can be developed
> for it ?
>
> I've seen that a virtual File System has been developed. What for ? As no
> applications are developed specifically for JDistro, what interest ? If we
> use KNOPPIX, then it will take care of that itself, no need to do it...
>
> I think now it will bring more visibility to the project if we say in the
> website, our goal is to ... For that, we need to progress in ... and ...,
> we are currently working on .... etc. We need to focus on something in
> order to motivate people !
>
> What do you think ?
>
>   G?rard

[Scope-dev] Re: [jdistro-devel] Where do we go ?

[Guillaume D. <gui...@de...>]

Hello,

Christian Heller wrote:
> (Again a cross-posting ... but this is an important integration issue.
> I think we all develop and develop but inter-communicate too little.)

I agree.

> JDistro:
> - Did you try to build on JOS or to integrate it into your distribution?

No. I'm very interesting in a JavaOS. But right now, the focus in on two 
subprojects: the desktop and the launcher.

> All:
> - Do you try to steadily improve your architectures ?

Yes. But the constraints are quite strong. We want JDistro:
- to run on a standard JVM
- to run java applications unchanged
- to be swing-based

 > and to be open for new, effective ideas?

Yes. But I believe more in new, effective code. Software ideas are great 
only when they are practical and implemented ;-)

> - Are you fixed on Java only or open to replace it wherever it turns
> out not to be the ideal solution?

I'm fixed on it. Not on the language itself but on the platform. The JVM 
and the java APIs are the fundation of the project. Considering that, 
the language itself can be changed, using jython, jruby, smalleiffel, ...

> Concrete for JDistro:
> - If the Desktop runs fine as it is, then just release your 1.0!

My personnal opinion is it's not ready to be tagged as 1.0. There is 
still some modules to improve. Maybe this year.

> - Knoppix is a good idea; I think JOS was thinking about the same?

Yes. If someone is interrested, it would be great to build a knoppix 
with a JRE, JDistro and a set of apps.

> - You mention a "JDistro library so that applications can be developed"
> which sounds much like an "application framework" to me.

Right now, there isn't. Our application framework is the standard java APIs.

> - The "virtual File System" is truely a duplication effort since JOS
> as potential operating system has developed one, too.

Any link about it ? When I searched the web about VFS written in Java, I 
found a few but none was transparent. All require to change your app. 
That's why I started the JDistro one.

Guillaume

[Scope-dev] Re: [jdistro-devel] Where do we go ?

[Christian H. <chr...@tu...>]

Hi Gerard, Hi Guillaume,

just some answers:

[Gerard]
> You mean the OS comes with a Framework for development ? Again, Java has

Yes. All systems/ applications (like Humans) are equal in that they have
input, memorizing, processing and output functionality. Operating systems
are not different from a small application here. Both can build on the same
framework.

> all this, even if it can (must) be refined by libraries like Scope (I can't
> find it, where is it ?)

http://scope.sourceforge.net/
http://sourceforge.net/projects/scope/

[...]
> quite well with other projects. Maybe it would be interesting working with
> the JOS Projet, but it seems to be dead since end of 2001. That was a

Many people (including me) seemed to think so due to the mixed website,
lack of its update and all the JOS-subprojects. Then I found out that
the list is alive! Mostly, Gilbert is working on the sources as it seems.
For sure, he can give you more insight on the current state.

> project I used to be interested in... For the MVC framework, it's not in
> the aim of JDistro to be a full featured framework, just to run most of
> java applications. If we can provide some simple yet powerfull API to the
[...]

But JDistro itself could base on a framework to:
- avoid writing redundant code
- get a clearer architecture
- unidirectional dependencies (security)
... and still be able to run standard java applications.

> Anyway, comments are welcomed and we would appreciate some compatibility
> checks with any library. I.E. if someone says to us, please make this
> application or framework work in JDistro, I think we will do it. Maybe
> that's the way collaboration can be built upon.

That's great.

[Guillaume]

> > - Are you fixed on Java only or open to replace it wherever it turns
> > out not to be the ideal solution?
>
> I'm fixed on it. Not on the language itself but on the platform. The JVM
> and the java APIs are the fundation of the project. Considering that,
> the language itself can be changed, using jython, jruby, smalleiffel, ...

That is good. Recent mails in the JOS list were about design faults in
the JVM and JDK and suggested a better architecture. (For details,
ask there. I'm not into it.)
If JOS provided a JVM with different design that were still able to run
standard Java applications -- couldn't JDistro use it then?

> > - Knoppix is a good idea; I think JOS was thinking about the same?
>
> Yes. If someone is interrested, it would be great to build a knoppix
> with a JRE, JDistro and a set of apps.

Just today, the http://www.gnumed.org /.net project announced the
availability of a Knoppix-based CD with GNUmed on it:
http://www.openmed.org/html/modules.php?op=modload&name=News&file=article&sid=64
The ISO image under:
http://marvin.ba-loerrach.de/gnumed.iso

It took them several months to accomplish this and I guess we could
learn a lot from them. See also the "Knoppix-Remastering how2":
http://www.linuxtag.org/cgi-bin/yabb/YaBB.pl?board=knoppix-en;action=display;num=1029373150

I am sure you could ask <seb...@gm...> or <ti...@de...>
for help.

> > - The "virtual File System" is truely a duplication effort since JOS
> > as potential operating system has developed one, too.
>
> Any link about it ? When I searched the web about VFS written in Java, I
> found a few but none was transparent. All require to change your app.
> That's why I started the JDistro one.

Uhh, I'm afraid I might have been "leaning out of the window" too much
here -- means: I don't even know what a "transparent" VFS is.
However, I had seen a package "filesystem" somewhere in the JOS sources.

Best Regards,
Christian

Re: [Scope-dev] Re: [jdistro-devel] Where do we go ?

[Ludovic C. <lud...@ho...>]

Christian Heller wrote:

>Scope:
>- It seems to be very quiet in your mailing list. Did you all give up
>or continue to contribute to the Free Software Community?
>- Did you encounter design problems in Scope? If yes, then which?
>What can we all learn from it?
>- If the design turned out to be incomplete: What should be changed?
>- If the project has stopped: Did you try to find a similar project to
>bring in your knowledge and continue your work?
>  
>
The mailing list is quiet for now but the development is active. In fact 
we are now working on a major 2.0 release with a lot of refactoring in 
the core classes.
Yes, there were design problems in Scope 1.0 ;-(
First, the public API was starting to grow a lot after i tried to add 
validation and security in the framework. Then, to add this 
functionality, i had to mess up a lot in the internals of Scope, because 
there was no plugable design.
What i have learnt from this is:
- to build a framework, think about the user: try to hide as much 
complexity from him, and put in the public API only what he needs to 
actually use, and put the classes used internally in a separate 
framework-impl.jar
- before adding new functionality, think about your object model, and 
redesign it if necessary to have a plugable architecture in the areas 
where you want to add the functionality
- use the XP best practices: write only the code that you need right 
now, write the unit tests before writing the classes (or at the same 
time), refactor early and often, write stories to define clearly what to 
do next.
and i din't find a project like scope that provides a MVC framework for 
Swing applications (except the framework in ResMedicinae that forked 
from Scope 1.0).

Ludovic Claude