GO!
hello folks.
It is with great excitement that I announce the beginning of the end,
project Game Over or project GO!
I was reading "Application Development Trends" and much to my surprise
stumbled across the following sentences "EJB may become the next test-bed
for Open Source software technology". It then goes on to say that "Most
vendors [...] may eventually become concerned that [...] "open source"
alternatives will become as potent in the applications server space as Linux
has in the OS space". It goes on to name open source companies, the first
one being Telkel.
You bet your sweet a*s, was my immediate reaction.
It is my personal belief as many of you know that this model we are
pursuing, one where we share the infrastructure to build our respective
business succesfully on top is the wave of the future. Something truly
exciting as it is a real mix of business and open source. Yes we are open
source, yes we embrace the GPL yet at the same time, yes we are here to do
business.
Project Game Over is a very simple vision. Actually I can't really call it
a vision, it is more a "test bed" as the author (Mr Jack Vaughan) states.
It is quite an insight to say that the next challenge for the open source
community is the application server. But for us it is mere excecution that
is ahead of us. Linux was a pure OS play, Apache took every one by surprise
on the server while everyone was fighting on the desktop, the Application
server is another story altogether.
Here is a field where many companies have invested many dollars and lives on
selling the license, yet here is a technology that is clearly infrastructure
and in our eyes as necessary as an IP address and as fit for the open source
development as the modular OS is. We are pursuing the webOS, like so many
other vendors, but our business plan and technology vision calls for an
openness of the infrastructure.
As all these startup heads and individual contractors, but also students and
academics join the jboss organization it seems to me that we have quite an
amazing organization in place, one that whose power we yet not fully
measure. See, jboss 1.0 was complete in terms of code, but was hard to get
into in terms of contributions. Following Linux 2.0' modular design we have
adapted the "divide to reign" philosophy and isolated the different parts of
the container. Thanks to Rickard Oberg and his revolutionary design, we now
have the kernel and the interceptors, the plugins, the different extensions.
All clearly layed out for integration.
Now folks that have just minutes to contribute can do so. Int the past few
weeks we have seen the addition of the pools, the container BMP, and the
SOAP proposal. Each and everyone of these is a fairly high level
contribution in itself. To us it proves that the design is clean and
provides a clean slate and starting point for many talented developers eager
to jump in jboss 2.0. It has shown something deeper as well, it has shown
that the open source development of this GPL application server, the jboss
server, can scale... a chain reaction really, that is what our early
measurements have shown, that this open source group can achieve critical
mass and that a chain reaction will work. We believe we can show yet
another success of the open source model, we believe we have everything it
takes to achieve that success.
In short the technology and the people are in place. This bomb is ready to
explode. So all we need to do is light the fuse, all we need to do is say
GO! This effort is very simple. We will try to complete the best
g*SDF@@#@!~! application server in the market. We will integrate with
Tomcat and catalina when it comes out to provide the full J2EE API
implementation in open source. We will focus on finishing jboss 2.0 with
EJB 1.1 compliance and move ASAP towards EJB 2.0 compliance. We will
release jboss 3.0 when we reach EJB 2.0.
So this is a call for help as well as a call to arms. Project Game Over
needs you, it needs your time, WE NEED YOU!. We will not finish this
project without you, your active involvement, your contributions, your code,
your patches, your documentation but most importantly your buzz. I want you
to be aware of the power we represent as a group and the buzz this
generates. I took full measure of the power of the buzz at JavaOne, jboss
is where the buzz is at and we must keep it this way. I want you to realize
that if each and everyone of you mentions jboss twice in one week, wherever,
on a mailing list at a dinner with friends, by 6 weeks we will have covered
the industry by and large. There is no little contributions in the open
source world just saying the word "jboss" is helping us, but you can also
code, since even the smallest project we have identified (couple of hours)
will help us greatly. Again just a good word somewhere in the press will
help.
At this point, what we have done is parcel out the work to be done. They
are not going down too much into the detail, just flesh out where people can
take ownership. Some things will take you a couple of hours (the 1.1 todos
in the code) some will take weeks/month
Here is how it works
1- pick something you like, you are interested in or you have already
done... heck even something to learn.
2- send us an email at [EMAIL PROTECTED] we will route people to
projects
3- we will assign team leaders to the groups and you should feel free to own
the project as you see fit
4- We will create a bugzilla entry for each and every little project
5- we will provide web based visual feedback on the progress of the
projects.
OK, in no particular order let's move on to the list of Todos, also I
realize that many of these plugins or little projects already have people
assigned to them. Don't worry just resend us a little mail saying "I am
already on this" and we will integrate all the information. This is sort of
a "census" of what is to be done and who is doing what... just so that we
get a good picture of what is happening right now and move on from here.
EJB Container
=============
* Develop a LRU cache for EntityBeans. Start with the random passivating
cache and modify as needed. O(1) operations are required. See EJBoss 1.0
discussions for reference.
* Develop a LRU cache for Stateful SessionBeans. See above.
* Develop transaction manager. Extend/Rewrite current JTA implementation.
* Develop SOAP distribution protocol plugin.
* Develop RMI/IIOP distribution protocol plugin.
* Multiple NYI in jboss2.0 code!!!!!
Persistence
===========
* Develop a file persistence manager for Stateful Session Beans. Already
partially done.
* Finish BMP persistence manager for EntityBeans. Partially done.
* Rework JAWS for flexibility. Only do EJB 1.1 CMP for now.
* Begin doing requirement and design thinking for EJB 2.0 CMP plugin.
* Integrate Cocobase, Castor, Ozone (see below)
Security
========
* Container implementation of interceptor for security
* User transparent propagation of credentials
* Integration/unification with Tomcat security development
* JAAS integration
* Dynamic API for user management in jboss
GUI
===
* Add EJX plugin for container configurations.
* Add EJX plugin for JDBC type mappings.
* Develop a GUI for administering jBoss through the RMI/JMX adaptor.
3rd party Integration
=====================
* CocoBase - Develop a plugin for CocoBase CMP support (www.thoughtinc.com).
Plugin interface is available.
* KAWA - Develop a plugin for integration with the KAWA IDE from TekTools
(www.tek-tools.com). Plugin interface is available.
* NetBeans - Develop a plugin for integration with the NetBeans IDE
(www.netbeans.org). Plugin interface is available.
* Castor - Evaluate the castor services for integration in jboss (castor
JDO).
* Ozone - as alternate CMP schema evaluate OODBMS. Design plugin interface
for Ozone.
JMX
===
* Develop an implementation of a JMX agent (MBeanServer). (use JMX classes
from JDMK 4.1)
* Develop a server configuration strategy to be integrated in the JMX agent.
Must be XML-based. Must support federated configuration for
"future-compliance" with clustering.
* Evaluate "distribution" strategies for jar management and delivery to
nodes.
* Rework logging. Divide logging into message logging and log message
delivery. Integrate JMS delivery.
* Develop a RMI adaptor for JMX.
* Develop a scriptable commandline tool for administering jBoss through the
RMI/JMX adaptor.
Clustering
==========
* Begin with requirements and design specifications of clustering to be
included in jBoss 3.0. Make sure that jBoss 2.0 is "future-compliant" in
terms of infrastructure. Identify parts of the server/container that
would be affected. Do prototyping of cluster management facilities. Use the
"Ecological clustering" draft paper (was "Biological clustering") as a base.
Think big. Think different.
* Begin identifying algorithms for implementing clustered services. Assume
that cluster management is taken care of.
* In particular Design and implement the caching algorythm for the clustered
server.
Documentation
=============
* Expand/update general docs about jBoss
* Add tutorials on how to deploy beans
* Fix all examples
Testing and Stability
=====================
* Create testsuite framework, and begin implementing testcases. Accent
should be put on automated test suites such as JUnit. API compliance and
nightly build should be taken into account.
Performance
===========
* Identify and eliminate bottlenecks in the design and implementation, this
is cross projects.
* Provide benchmark numbers on optimized version and standard version of
jboss.
* Provide benchmark numbers on integrated J2EE stacks (tomcat+jboss)
EJB 2.0
=======
* CMP (see above) design and implement first version of dependent and helper
beans
* EQL, integrate the Query Language from 2.0 specification
* Message Beans. This is a container effort, provide the container code and
tools to read XML files and deploy the message beans in addition to more
traditional Session and Entity
* RMI-IIOP integration for container invocation (see above)
We have probably forgotten about a thousand things. We are also very
interested in you proposing additional projects you want to take on.
Now is the time to join, now is the time to prove that a J2EE infrastructure
is indeed an open source endeavour, now is the time to join "project Game
Over".
Go go go!
kind regards
marc fleury
----------------------------------------------------------------
Visit Telkel at JavaOne(SM), Sun's 2000 Worldwide Java Developer
Conference(SM)
June 5-9, 2000 - Moscone Center - San Francisco
