KSL: announcing new project Simeuluë and asking some initial questions

Thorsten van Ellen thorsten at vanellen.de
Tue Mar 3 14:53:31 PST 2009

Hi all!

*** Announcement ***

According to the invitation for announcements on the KSL home page, I announce 
the project Simeuluë:

I plan to start an open source project as part of my dissertation. The project 
is called Simeuluë like the island in the near of Java. In my dissertation I 
found situations (that I call contingencies) that precede error situations and 
the target is to enable Java to detect and handle contingencies before they 
become errors.

This is a new concept that is also valid for other languages. The raw concept is 
clear, some scientific publications already have been successful. The open 
source implementation will follow. A scientific evaluation is planned and at the 
same time the implementation can also be tested publicly. Finally, if everything 
is successful, I want to propose the implementation as JSR. So the JSR is the 
final but optional target. Of course, I can not guarantuee, that the JSR or any 
other part will be successful.

*** Questions ***

Before I really can start with the open source project some questions have to be 

* Must I really build a complete OpenJDK repository for my project with about 
30.000 files or which parts are relevant for me? I plan to extend the syntax, 
compiler and optionally the standard library. In the worst case even the byte 
code and/or the jvm would have to be extended.

* How can i prepare best for a simple future "JSR-patch" of the official JDK?

* Which JDK would be the best starting point? 6 or 7? Which Build?

* Where can I or should I start the open source project?
E.g. at dev.java.net or openjdk.java.net or is there any other more reasonable 
place? Part of Kijaro project (SVN) or own project (SVN on dev.java.net or 
mercurial on OpenJDK)?

I fear that I still do not understand all relationships and consequences of the 
choice. As far as i understand:

Advantages of kijaro:
* early "code cooperation" of simeulue with other projects
* "automatic" updates of official JDK versions(?)
Disadvantages of kijaro:
* no free choice of JDK versions

Advantage of OpenJDK:
* easier merge with mercurial of Simeulue-"patch" with official JDK(?)

Advantage of separate project on dev.java.net:
* free choice of JDK version

Are there other important advantages/disadvantages?

What license is appropriate for the open source project to be able to propose a 
JSR finally? GPL v2?

Do I need to sign a SCA? This is no problem, I only need to know, if I have to 
do that and when.

When is the best time to start with the JSR? Early, to give all participants a 
chance to influence the design? Late, to show that it already works? How can I 
find a good expert group for the JSR? I could imagine some known people that are 
"active" in the exception handling domain, that do not know me. Simply 
ask/"motivate" some of them? Or how does an expert group usually constitute?

I already started with a page on dev.java.net:
There you can find more information about the open source project and some 
scientific articles.

Please include a link on the KSL home page to my project:
Simeuluë: Java/Eclipse Extensions to Avoid/Repair Errors before Runtime 

Do you have some more tips?

Forum for further discussions:

Best regards

Thorsten van Ellen

More information about the compiler-dev mailing list