NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/lang/c++-part2/11175

Internet Message Format | 1996-08-05 | 3.1 KB

Path: news.dal.ca!Skye!kelley
From: [e-mail] (Dan Kelley)
Newsgroups: comp.lang.c++
Subject: Re: Will JAVA kill C++?
Date: 13 Mar 1996 01:13:35 GMT
Organization: Dalhousie University
Message-ID: <4i57fv$[e-mail]>
References: <[e-mail]> <4hp18v$[e-mail]> <4hq2j6$[e-mail]>
NNTP-Posting-Host: skye.phys.ocean.dal.ca
X-Newsreader: TIN [version 1.2 PL2]

Tom Payne ([e-mail]) wrote:

: When one is interested in speed, one should avoid pointers (as
: does. Fortran), since pointers kill opportunities for optimization.

As a scientist who once walked a FORTRAN highway daily, who took an
off-ramp to c++/perl/matlab/gri, I have to take the bait thus
offered.

Am I to suffer yet another language shift?  I have taken some time to
learn C++, or come close to learning it, and now this thread makes me
fear another computer language maelstrom.  I know java and C++ share
roots, but as a resident of a bilingual country I know that similar
language roots don't always translate into ease of use.

I share a lot of code with other scientists, and I've been
disappointed many times at how hard it is to get textbook c++ code
working.  Vendor compilers suffer differences in template
instantiation, forcing me to use all sorts of conditional compilation
messiness in my C++ code and my gnu autoconf scripts.  And if g++
(yes, scientists use this more than all you professional programmers
might know) is any indication, the textbook code is still a long way
from being workable in practice, what with continual development of
template/exception/... mechanisms.  As a scientist (and here again I
must apologize to you professional folks who have time to keep up with
the trends) the changing ground underneath my feet is distracting from
the real task of my research.  I don't long for fortran, but I do long
for a language that will work on many machines, and which I can
reasonably expect to work a year from now when I recompile with newer
compilers.

So my question to the experts on this newsgroup is simply this:
compared to C++, is java code really, in practice, transferrable from
one machine to another?  And will java code that works this month
likely work a year from now?  If so, then my vote on this thread is
that java does, in fact, have the potential to kill C++.  For my
experience as a non-dedicated and non-professional programmer may
apply even to the experts who will read this: my time is better spent
investing in code than in tweaking it from machine to machine and year
to year.  All of this amounts to nothing more than expecting textbook
code to actually work on an arbitrary machine.

PS: thanks to any of you who read through to the end of this.  I
realize my comments may be out of place on such a technical group and
such a detailed thread.  I offer my thoughts only in hopes of adding a
reflection from angle.  Thanks.

--
Dan E. Kelley                   internet:         mailto:[e-mail]
Oceanography Department         phone:                       [phone]
Dalhousie University            fax:                         [phone]
Halifax, NS, CANADA, B3H 4J1    WWW:  http://www.phys.ocean.dal.ca/~kelley