NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/alt/msdos/programm/3005

Internet Message Format | 1992-12-24 | 8.2 KB

Xref: sparky alt.msdos.programmer:3005 comp.os.msdos.programmer:11608 comp.sys.ibm.pc.programmer:670 comp.lang.c++:18392 comp.lang.c:18867
Newsgroups: alt.msdos.programmer,comp.os.msdos.programmer,comp.sys.ibm.pc.programmer,comp.lang.c++,comp.lang.c
Path: sparky!uunet!grebyn!daily!malak
From: [e-mail] (Michael Malak)
Subject: Re: Newbie Wants Advice on C-Programming
Message-ID: <[e-mail]>
Organization: Grebyn Timesharing
References: <[e-mail]>
Date: Thu, 24 Dec 1992 17:23:33 GMT
Lines: 164

In article <[e-mail]> [e-mail] (Jon Noring) writes:
>Hello,
>
>1) There seems to exist C, as well as C++, and possibly other variants.  What
>   are the differences between these?  Which should we learn first?

In my opinion, there is no fast track to either C or C++.  I feel the
following must be learned in order:
   1) Pascal, with two semesters experience
   2) Assembly, learned anywhere, with tinkering on your own
   3) C, 2-3 years min. anywhere (college, real world, or tinkering)
After that you can learn and appreciate C++.  You need to first learn
structured programming and algorithms, most importantly, linked lists
and trees.  Since C is tightly coupled to assembly in many ways (C is a
shorthand for assembly, and C integrates easily with assembly modules),
you need to known assembly to understand what is going on in your C
program.  The after you know C, it takes _years_ of experience to really
know it.  After 6.5 years, I'm still learning.  After 5 years, I was
doing object-oriented programming in C (not C++) without really
realizing it.  It was at that point that I could appreciate the
syntactic shorthand of C++ for object-oriented programming.  I've never
programmed in C++, BTW, due to the lack of a really good mainstream
32-bit C++ compiler for MS-DOS.

To summarize, C programs should appear structured like Pascal programs,
but are really shorthand for assembly programs.  C++ facilitates
object-oriented programming, but you need experience when to use it
because you can too easily introduce inefficiencies.

>
>2) We want to do most of our programming on our 386-33 PC running MSDOS 5.0.
>   What C-type compilers exist for the PC?  Which ones are better?  How much
>   do they cost and where can one get them at a discount?

The first question is 16-bit or 32-bit.  16-bit compilers generate
native MS-DOS programs and MS Windows 3.x programs.  32-bit compilers
generate DOS-extended programs (which appear to run just like native
MS-DOS programs) and, in the case of Microsoft's C compiler, Windows NT
programs.

The best 16-bit compiler ever made is unfortunately no longer available.
It was Microsoft C 6.0.  I should qualify "best"; I mean in terms of
professional MS-DOS development.  Although I have little experience with
Borland C compilers, I would have to recommend it to you for the
following reasons (if you want a 16-bit compiler):

1) Beginners tend to love Borland's "integrated development environment",
   where I presume you can edit, compile, link, and debug from what
   appears to be the same program.

2) Microsoft C++ 7.0 requires the finicky 386-MAX memory manager for
   compiling and linking under DOS, thus making it unattractive for
   creating DOS programs.

3) Borland has aggressively added support for creating Windows programs,
   so much so that it is unclear which is better for Windows development
   (typically programmers will stick with the "official" C compiler no
   matter how bad it is, because of compatibility and compiler bug
   issues).

For 32-bit development, the Intel C Code Builder is in my opinion the
best for creating DOS programs since it can link in MASM 6.0 modules.
MASM 6.0 is the best assembler I have ever seen for any computer.  It
allows you to create very structured and readable assembly code, and
seamlessly shares data and functions with C.  Note, however, that it is
just a C compiler, not a C++ compiler.  As I said above, there are no
really good mainstream 32-bit C++ compilers.  I don't feel Zortech C++
qualifies because it doesn't come with a 32-bit C++ debugger, I keep
hearing bad things about it, and it uses a 286 DOS extender with a 16MB
memory barrier.  The memory barrier on a 386 DOS extender, such as the
one built-in the the Intel Code Builder, is 4GB.

For 32-bit Windows development, the only choice as far as I know for
Windows NT is the special version of Microsoft C++ 7.0 that comes with
Windows NT beta developer's kit.  There are 32-bit C compilers for
Windows 3.x, but I wouldn't recommend them since they are not
mainstream.

To me, "mainstream" is important because it means a higher quality
product and better support.  It means more complete and more
professionally done documentation.  It means more complete and wider
beta testing.  It means larger resources to fix the bugs that you find
and report in their compiler.

In any case, you should get MASM 6.0, which lists for $150.  It is both
a 16-bit and a 32-bit assembler and can link with Microsoft, Borland,
and Intel C compilers.

The Intel C Code Builder lists for $595.  I'm not sure of the prices for
Microsoft C++ and Borland C++, but I imagine they are $300 - $500.

>
>3) How robust is C compared to Fortran for purely number crunching, scientific
>   type applications on the PC?

C does not do range-checking on array indices.  This is one way that C
is coupled to assembly: a[i] means _exactly_ "add a to i and
dereference".  In fact a[i] and i[a] are synonyms!  If you want to incur
the overhead of range-checking, you could easily do so in C++.

One reason that FORTRAN is associated with good number crunching
capabilities is that there exist out there optimized, proven math
libraries.  I would guess that that is not so much of an issue anymore,
that math libraries are available and/or have been ported to C, but I've
never done scientific number crunching in FORTRAN, so I don't know what
I'm missing.

>
>4) What books would you recommend in learning C (my wife might want to take
>   classes - I do better teaching myself)?  And of course, since I will be
>   doing programming on a PC, what books would you recommend that are
>   specific to that platform, especially w.r.t. memory management as well
>   as graphics?

You should always start out with the official books:

For C, it is _The_C_Programming_Langauge_ 2nd ed. by Kernighan & Ritchie
For C++, it is _The_C++_Programming_Language 2nd ed. by Stroustrup
For assembly, it is the programmer's reference manual publsihed by the
    chip's manufacturer.  I would recommend the i486 programmer's
    reference manual published by Intel.  The invaluable quick reference
    guide that comes with MASM 6.0 tells you which instructions are
    available on which processor(8088, 286, 386, and 486).

For each language, you may optionally buy a more tutorial or descriptive
book.  The books listed above are more reference books than tutorials.
Although they do start with the basis, and then build on them, they do
so in a math book kind of way.  Just go to a computer book store and
thumb through some of the books and buy the one(s) that appear to teach
the way you like to learn.  Use the reference books to verify the
tutorial books, and to provide complete information.

As for memory management, it is a non-issue in the 32-bit world, which
is the way the world is going.  The kluges invented for the 16-bit
DOS world, expanded memory (EMS) and extended memory (XMS) are pretty
technical and not necessary for either 16-bit Windows or 32-bit DOS
extenders.

As for graphics, you'll need to use Windows calls to do graphics under
Windows.  I've never done any programming under MS Windows (I wrote my
own windowing system since MS Windows was too slow for the real-time
displays required by my current project), so I don't know any books to
recommend.  To do graphics under DOS, you _might_ find the library calls
that come with the C compiler sufficient.  Most people just write
directly to the VGA registers.  The standard, but _poorly_ written,
guide is called something like _Programming_EGA/VGA_, 2nd edition by
Ferarro.

>
>
>Much thanks for your help/advice.
>
>Jon Noring
>
>
>(p.s., I looked for a FAQ about my questions, but did not happen to find one,
>so please no flames in the event there is a FAQ for this.)
>
>--
>
>Charter Member of the INFJ Club.

- Michael Malak, INTP