NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/lang/c++-part2/11292

Internet Message Format | 1996-08-05 | 3.8 KB

Path: ra.nrl.navy.mil!usenet
From: [e-mail] (Richard Pitre)
Newsgroups: comp.lang.c++
Subject: Re: Will JAVA kill C++?
Date: 13 Mar 1996 15:34:02 GMT
Organization: Naval Research Laboratory
Message-ID: <4i6ptb$[e-mail]>
References: <[e-mail]>
NNTP-Posting-Host: n5160d.nrl.navy.mil

In article <[e-mail]> Larry Baker <[e-mail]> writes:
> Adam Megacz wrote:
> > I'm not a Fortran buff, but I don't understand why C is faster
> > than C++. C++ supports all C features (except the void* crap).
> > If anything, C++'s type-checking allows more compiler
> > optimizations, making it a slightly faster language.
>
> If you code _exactly_ the same way using a C++ compiler, in
> essence using the C++ compiler to compile a "C" program, then
> you should see no performance difference on non-optimized code.
>
> C vs Fortran raises a few other interesting comparisons. All variables
> in FORTRAN are the equivalant of C "static" types - they aren't
> dynamically pushed/popped off of the stack, as with automatic C
> variables, or dynamically allocated using malloc/free, as they
> may be (even as automatic, stack-based instances) in C++.  Hence
> the C runtime semantics, and moreso the C++ runtime semantics, may
> incur some slight "background" overhead when compared to a much
> simpler language like FORTRAN.
>
> Larry Baker
> [e-mail]

I agree with what Larry is saying. A programmer, who has some sense of how C,
C++, and Fortran are typically implemented, can do everything efficiently in
C++ with the added advantage of classes. C++ gives you fine control over
efficiency and organization and the compromise between them.

IMHO C++ is a tool only for experienced and disciplined professionals. I think
that it is important for programmers to understand the motivational issues
behind the C++ features and the place to find that motivation is in
Stroustrup's excellent writings. Some people complain about the difficulty of
these books but I believe that much of the difficulty is inherent in the
subject i.e. C++ is complex and subtle to the point that people seem to have
difficulty in correctly implementing some of its more useful features. This
reading should be a sober experience for anyone making claims for C++ as an
ideal all purpose tool for the programming masses. The notion that less
experienced programmers will somehow confine themselves to a somewhat safe
subset of the language is a foolish notion. The popularity of C/C++ is a
serious problem not only for people who need other tools but perhaps for the
future of C/C++ itself.

Beginning and intermediate programmers should use something other than C++
until they understand what programming is really about. Pascal lineage
programming tools(and perhaps JAVA) are good alternatives for beginning,
intermediate and many professional programmers, although as Larry pointed out
these languages are largely non existent thanks to the gross over popularity of
C/C++. Of the Pascal lineage languages Ada95 is a real standard that seems to
have significant multiplatform support and there is some inkling of a
competitive compiler market. Ada also has some heavy duty annexes to do real
world things. Modula 3 may also be a practical choices for many.

After some experience with tamer languages and  after reading something on the
order of Stroustrup's books, a programmer can make a conscious choice to use
C++. Many programmers may conclude that, in spite of their experience, they do
not have the necessary discipline to code in C++. If every programmer could
make that kind of conscious decision then there would be healthier software out
there and there would be serious competitive alternatives to the supposed one
true tool. Dreamland, I know.

richard