Date: Tue, 20 Aug 1996 00:39:51 +0200 (MET DST)
Message-Id: <199608192239.AAA27751@dxmint.cern.ch>
Mime-Version: 1.0
From: purschke@axgsi5.cern.ch (Martin Purschke, GSI, CERN-PPE)
To: lueking
Subject: Java
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit


Hello Lee,

a quick answer for now, I'm a bit busy. As of next month I will be
at BNL and work for the PHENIX experiment, so we can continue the discussion 
then.

I did these Java demo to find out how far I could get (and how fast) 
with zero experience with Java (but long-term with C++ and OO in general)
and with a working C++ implementation of our analysis code. What you see
on my web page is the result of a rainy weekend's work. About half
of the time I spent on the histogram display class, since I couldn't find 
a good histogramming package in Java. The rest (converting the C++ code)
was painless; there were a few techniques I had to figure out, like substitute
subarray passing (no pointers in Java) and some more, but once I had it
this went fast. Then some file/stream/URL IO, and there it was. 

My impression is that Java has about the same capabilities as C++, but 
tries very hard to hide the details better than C++ does. I think that
a fortran programmer would be more comfortable writing a few Java snippets for
his/her analysis than doing it in C++. In addition, Java comes bundled with a
generic windowing system built in (the AWT, abstract windowing toolkit, which
shields the user from many details). The effort to "just open a window and
display something" is a snap compared to what it takes in C/C++. And Java has
all the OO power, it is not a toy. In general I think that Java will play a
very important role, all the more as today's CASE tools are able to produce C++
as well as Java (or some other, like SmallTalk) code, so the actual language
will become less important. Other than all other languages, Java has the
platform independence advantage, Java code can be excuted on any platform,
and even across the web. 

I am not aware of other people using Java for real work at CERN, but my web
page created a lot of network traffic, and I got a lot of comments. However I
think that the current attitude is "real men do it in C++, let's worry about
this later", and it will take some time for Java to gain general acceptance.

Ok, so far for now, I hope this helps a bit.

	Cheers,
		Martin