[Scope-dev] Collaboration [3]

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

----------  Forwarded Message  ----------

Subject: Re: [jos-general] Re: [jdistro-devel] Where do we go ?
Date: Wed, 02 Jul 2003 18:46:59 -0400
From: Gilbert Carl Herschberger II <gc...@mi...>
To: Christian Heller <chr...@tu...>

At 11:29 PM 7/2/03 +0200, you wrote:
>At first, one question: Is there a reason you didn't send your email
>to the lists? I didn't find any confidential things. On the contrary,
>it contains good and important proposals!

I didn't sent my e-mail directly to the mailing lists because I thought I
might be asking a lot of you. I did not know how you would feel about that.
Now that we agree something must be done, should you post my message (and
yours) on the various mailing lists? I think you might first do a little
research on the Debian effort. Gather the facts before presenting it to
others. If another project is already trying to do this, I would like to
know more about it.

I would like to install a Java-oriented--optimized and
enhanced--distribution of GNU/Linux. That would be very useful to me, too.
Is that a goal for Debian?

Am I speaking of a distribution project? Hmm, maybe I am. My very first
goal for the JOS Project was to provide the JOS Technical Edition, a
distribution of JOS-related stuff. I made it easier to find stuff. Why? I
didn't want others to have as much difficulty as I did. Certainly, a
distribution project could distribute a collection of "friendly"
applications that are designed to work together. As a super-project, we
might work to "improve" or "upgrade" existing Java applications so that
everything can work together. We can improve the out-of-box experience.

Then again, maybe I'm not. As you may have heard, most of the JOS software
does not work together. Isn't that sad? There are lots of little
incompatibilities.

But of course, all /my/ stuff works together. ;)

When distributing the files that other people made, I could do very little
to convince them to change their product so that it worked with other
products. I am convinced that this might be where Java technology suffers
the most. It is so easy to write a Java application that cannot run
alongside other Java applications. But I should be able to run all Java
applications side by side without such incompatibilities. We have an
amazing class loader mechanism that few people are putting to good use. We
have an amazing namespace mechanism so that different applications don't
need to use the same namespace.

I would like to help define what it takes to build, not just a Java
application that works, but a Java application that works with other Java
applications. Could this be such an opportunity? Is this the right time?

This is vaguely familiar. It reminds me of the RedTea project on
sourceforge.net that set out to build a Java equivalent of RPM, the RedHat
Package Manager. I haven't heard from them in a while. It was "red" because
of RedHat. It was "tea" in response to Java (coffee). The idea was that,
with an equivalent of RPM, Java products could be distributed similar to
the way products are distributed on Linux. The community wasn't ready for
yet another RPM. It seems that it might be better to determine how Java
applications might be tested for "compatibility" and help authors
distribute their products using RPM. Maybe distribution is a part of it.

Let's say you wanted the latest version of every Java product on
sourceforge.net. How would you do it? Get a distributor. A distributor
involves people from many projects. They ask authors to build an RPM for
each product. They mirrors the RPM in a central location, hopefully taking
care only to introduce stable combinations of products, giving you one-stop
shopping. And hopefully, taking care of any licensing issues. (1) Is every
Java product available as an RPM? (2) Has anyone volunteered to be a
distributor?

Wow! I didn't realize how strongly I felt about this. As it should be,
RedHat is an expert in the distribution of Linux, not Java. Who is an
expert in the distribution of Java products? Is anyone?

Thanks,