[Mono-list] Please comment on excellent Java/J2EE/EJB v. Mono/.NET article here...

Jason Watkins jason_watkins@pobox.com
Tue, 30 Oct 2001 04:47:48 -0800

Well, I have a few issues with what he says:

1.) MS Hailstorm, and sepcificly Passport is the *real* issue.

The CLI just happens to be the plumbing for this particular poison. What
everyone who fears MS.net should be doing right this very moment is figuring
out how to provide service providers and service users the same features as
Passport. They should develop it NOW and market it hard. Controlling and
verifying indentity is going to be key to the future we're trending toward.
Personally, I'd like to see an even playing feild of identity providers
using a seemless system. I'd like to see an appropriate body hold domain
over who is allowed entry into the top tier of these providers, and
continual review and certification.

I'm sure there are flaws with this scheme... it resembles the current DNS
system, and we all know how flawed that has become. I'd like to see other
people comment on just how a distributed identity system might work.

Convincing 3rd party service providers not to use passport exclusively is
going to be the true task the OS world should face. This means working with
lawmakers and regulatory bodies to make them aware that technology allows
systems far more secure, just and private than handing this off to an MS or
Oracle chaired federation.

2.) As a Java user and evangelist, he sees no problem with saying "we should
all just use java".

Why should you ever need VB, C#, C++, OCaml? Java is one size fits all!
Sure, a common type system is cute, but who needs it when Java's types are
more than adiquate!

Java is a fine piece of technology. It has good uses. In my most recent
corperate reincarnation, I specialized in servlet based CRM software. I
liked it quite a bit better than MS based competition.

But anyone who has uses STL or templates in c++, or needed to do big number
cruching in Fortran, or needed to write code with certifiable correctness
with ML languages knows what everyone here most likely knows: there is no
silver bullet for the task of programming. There is no one size fits all
programming language.

Forcing Java to be the only major language could hold back the future of
computing by at least a decade.

In the future that is comming we will attack programing tasks so utterly
stagering in thier complexity that we will simply give up all notion of
designing the entire imperitive solution. Even describing the fitness of
potencial solutions will be a difficult task.

Can you imagine generic programming grafted into Java?
Can you imagine highly distrubted systems in Java?
Can you imagine intelegent systems on the scale of even the average feline's
motor functions in Java?

Maybe you can defeat me on those 3 points, but I think you'll agree that as
the future comes into being, we will transition into new paradigms of
programming. Imperitive programming a la Von Neuman is here to stay, but the
frontier is constantly expanding, and these languages will be hard pressed
to make the journey.

Even the CLI already is showing it's limitations when you think about
implimenting Ruby, Python or the like in it.

Sun may be less evil than MS, but I still don't like there being any
monoculture. I like java, I like sun. I even like Bill Joy tho I disagree
with his rants. But I don't like one king of the hill... I'd like to see
more competition and interoperatility.

So anyhow, where I feel he's wrong is:
1.) Passport is the real problem, and pressure needs to be applied so that
this doesn't turn into fiasco so big people will wish for the days with
InterNIC bungled and abused their DNS rights.
2.) Java is not appropriate for all tasks. Neither is .net for that matter.
But what we do need is a loosely coupled platform where different language
technology's are allowed to interact as much as possible. .net doesn't
wholely solve that problem, but it definately steps in the right way. We
should not throw that baby out with the Passport bathwater.

jason watkins