Re: Jigsaw 2.0 !

Allow me to enter my $0.02 here.

First let me remind everyone of two very important benefits that I believe
we will see as a result of the changes Anselm has proposed:

1) The ability to use Jigsaw not only as a stand-alone Web server, but also
as an extension to _any_ Web server that implements the Java Servlet API.

2) A significant simplification of the process by which one writes
Resources that extend Jigsaw's functionality.  A lot of human overhead is
currently required in order to maintain the AttributeRegistry and each
Resource's Attribute array.  I believe the transition to JavaBeans will
make this extra work simply go away.

I think it makes a great deal of sense to transition Jigsaw in the ways
Anselm has described.  Jigsaw's serialization and "reflection" mechanism,
based on the attribute array, made a great deal of sense when the Java
Platform did not itself provide serialization and reflection.  To continue
development on a version of Jigsaw while based on the "outdated"
attribute-array-based system seems illogical, now that the language has
evolved to provide serialization and reflection in the standard API.  It
makes more sense to do what Anselm has proposed: make the changes that the
new API allows in the implementation of the Resource object model -- thus
greatly simplifying Jigsaw _and_ the process of extending it -- and then
continue development from there.

I think it's an excellent idea to release a current snapshot of Jigsaw as
1.0 final, and then immediately begin work on the next version, using the
revised and simplified implementation of the object model.  The proposed
changes are significant enough, in my view, to warrant a major version
change.

Mark Friedman, in his prior email on this thread, makes an excellent point
that "creeping design-itis" is a bad thing, and can infinitely delay
delivery of a final product.  However, I don't think that's what's going on
here.  Changes for future versions of the JDK are not going to be nearly as
sweeping as the changes that occurred between 1.0.2 and 1.1, especially in
terms of lower-level language issues such as reflection and serialization.
From all indications, based on things I heard and saw at JavaOne, it will
_not_ be the case that the next version of the JDK will cause another major
re-evaluation of Jigsaw's implementation, so I don't think things will
continue to "creep."  This is simply a matter of making sensible
conclusions about Jigsaw, based on experience and the evolution of the
underlying technology (Java).  It would be self-defeating to ignore these
issues.

We've all been working with Jigsaw as an "alpha" product.  Part of that
bargain is the understanding that significant changes can be expected to
occur as development proceeds.  Anselm has been, and continues to be, very
sensitive to users' needs for stability.  The changes Anselm has proposed
will benefit us all, and should be made sooner rather than later.  It's my
opinion that these changes will make any Jigsaw-based technology much
stronger in the long run.

Bring on 2.0!



+---------------------+-------------------------+---------------------+
|    Dave Makower     |    davemak@pencom.com   |   WWW Specialist    |
+---------------------+-------------------------+---------------------+
|     Co-author of "Java Programming Basics" (Henry Holt/MIS:Press)   |
|                 http://www.pencom.com/javabasics/                   |
+----------------------------------+----------------------------------+
|   Pencom Web Works               |   (212) 513-7777   voice         |
|   Pencom Systems Inc.            |   (212) 513-1014   fax           |
|   40 Fulton St.                  |                                  |
|   New York, NY  10038            |   http://www.pencom.com/         |
+----------------------------------+----------------------------------+

Received on Wednesday, 16 April 1997 16:48:01 UTC