NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/lang/c++-part2/13471

Internet Message Format | 1996-08-05 | 7.8 KB

Path: scoop.eco.twg.com!usenet
From: [e-mail] (This space intentionally left blank)
Newsgroups: alt.computer.consultants,comp.edu,comp.lang.basic.misc,comp.lang.c++,comp.lang.misc,comp.lang.pascal.borland,comp.lang.pascal.delphi.misc,comp.misc,comp.os.msdos.programmer,comp.os.os2.programmer.misc,comp.programming
Subject: Re: Info on being a Systems Programmer/Analyst?
Date: 25 Mar 1996 21:20:27 GMT
Organization: The Wollongong Group
Message-ID: <4j72mr$[e-mail]>
References: <4itd85$[e-mail]> <[e-mail]>
    <4iv232$[e-mail]> <4ivirl$[e-mail]>
NNTP-Posting-Host: jake.eco.twg.com

In article <4ivirl$[e-mail]>, [e-mail]
 writes:
>In <4iv232$[e-mail]>, [e-mail] (This space intentionally left blank) writes:
>>A SYSTEM PROGRAMER does the same sort of things as an applications
>>programer, but closer to the hardware (i.e device drivers, operating system
>>alterations, new system services, etc.).
>
>They used to... But today device drivers are shipped with the device itself,
>operating system alterations are as rare as the dodo birds, and if companies
>want new system services, they just buy a package. It is simply not worth
>for most companies to pay someone to perform those 'alterations'. There
>are quite a few companies that provide solutions for special needs.

Ummmm....excuse me, but who do you think writes the stuff that gets shipped
with the device, or provided by the OS provider or makes the packages you
talk of?  System Programmers.

You are right that most companies have no need for system programers these
days, but the ones who do, need them to stay in business becausr their
business is providing system software.  The place I work still writes device
drivers because that is part of the product we sell to customers.

Most "end-user" companies (the majority of companies these days) have no
need for system programmers.  They need system managers/administrators and
they may need application programmers (unless they are in a "vertical
market" and the specialized stuff they need is commercially available or
contract out all their development), but the days of every installation
needing a system programmer are gone.  When the system crashes and takes a
dump, you don't call in your own personal system programmer these days, you
call the support number for the folks you bought the software from, send
them the dump and *they* call in their own personal system programmer to
look at it (we're assuming real computers here, not "press the reset button
if it locks up" toy machines ;^).

>>These beliefs are backed up by looking at the content of Digital Equipment's
>>courses.  The System Programing course has to do with system services, ASTs,
>>access to kernal memory, etc. while the System Management course has to do
>>with insalling software, backing up disks, mounting tapes, tuning for
>>performance and allocating resources (accounts, disk quotas, privileges,
>>protections, etc.).
>
>Do not be misled by the courses, they have their inertia... For example, I do
>not know for sure what kind of education you will get in a university if you
>want to be a programmer, but I am pretty sure that in most (if not all)
>countries they still teach you things like 'how to solve a differential equations',
>and they will go deep into higher mathematics, but will scratch the surface
>when it comes to computer-related theoritical issues.

Can't say what they're teaching these days, but when I graduated with a
computer science degree 16 years ago, they taught a whole bunch of things,
among them differential equations, matrix math, boolean algebra, finite
state machines, calculating algorithm efficiency, probability and
statistics, numerical methods, data structures, graph theory, a number of
languages (if you hadn't written code in at least 6 or 7 you weren't
graduating...I think my total was 9, two of them assemblers (one mainframe,
one micro)), operating system and compiler design, and how to write programs
that don't break at the first bad input.

One instructor was famous for disobeying program prompts.  If the program
said, "Enter a number between 1 and 10" he would enter something like
"Billy" and if the program blew up, you flunked the assignment.  Being late
with an assignment cost you one letter grade a day.  Writing code that met
the requirements of the assignment got you a "C"...after all, doing what was
asked is about average. :^)  If you wanted a higher grade you had to do
something extra, or several something extras.  For example, if the
assignment was to write an assembler for the DECsystem-10, and you did that,
and it ran properly, you got a "C".  If you did it as a one-pass assembler
or generated a nice listing file with octal (36-bit machine) object code
shown, you got a "B".  Do a one-pass that also generates a nice listing and
you might get an "A".  Of course, if you had bugs in the extras, it could
sink you to a "D" or an "F", so make sure the basics are bulletproof before
getting fancy... Sound "real-world" enough for you?  ;^)  (BTW - this
professor was the only one on the faculty with real programming experience
as a working programmer rather than an academic...and it showed!  Dr.
Dunning...are you out there somewhere?  If so, THANK YOU!!!)

>My approach to the original question was from a practical point of view.
>I simply did not want to misinform the person who asked the question by
>telling him something that according to my experience belongs much more
>to the past than to the present (and possibly future).

There will always be a need for system programmers (at least as long as
there's a need for programmers at all...we'll see how the AI research
goes... ;^), but the total number required will never be as large as it was
when you had to have at least one for every mainframe...at least as a
percentage function of installed machines.  The larger overall software
market today may actually employ more than the relatively small market of
mainframe days...

>There must be ACTIVE systems programmers who read some of these
>newsgroups and can tell us how much 'administrative' work they do and
>how much 'system alterations'. If you are out there, please speak up and
>make me stand corrected.

The ones we have here spend 80% of their time on coding, debugging and
researching requirements for new code.  The remainder is mostly meetings,
staying current and talking to customers with really tough problems.  The
"system manager" type work is a non-zero, but tiny fraction of their time,
and it tends to come in chunks as we get a new beta version of an OS to set
up and test, or new hardware comes in and has to be configured and put into
use.  Of course, our "users" tend to be other programmers, who can usually
do thier own changes as needed (if I need a test account set up, I just do
it myself and I'm more of an application programmer than a system one,
though I've done both, and system management as well, in the past.

The kind of environment has more to do with the type of programmers and thier
duties than anything else.  A production environment is different from a
software development environment is different from a research environment.
I've seen all of these at one time or another and all have their patterns.

     -- Mike "system programmers are not an extinct breed" Bartman --

==============================================================================
| I didn't really say all the things that I said.  You probably didn't read  |
| what you thought you read.  Statistics show that this whole thing is more  |
| than likely just a hideous misunderstanding.                     |
==============================================================================

==============================================================================
Prosecutors will be violated
------------------------------------------------------------------------------