NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/comp/edu/1488

Internet Message Format | 1992-09-02 | 5.0 KB

Xref: sparky comp.edu:1488 comp.lang.fortran:3382 comp.lang.misc:2893 comp.arch:9175 sci.math:10882
Newsgroups: comp.edu,comp.lang.fortran,comp.lang.misc,comp.arch,sci.math
Path: sparky!uunet!ftpbox!motsrd!white!sapphire.rtsg.mot.com!declrckd
From: [e-mail] (Dan J. Declerck)
Subject: Re: Scientists as Programmers (was Re: Small Language Wanted)
Message-ID: <[e-mail]>
Sender: [e-mail]
Nntp-Posting-Host: marble
Organization: Motorola Inc., Cellular Infrastructure Group
References: <[e-mail]> <[e-mail]> <[e-mail]>
Date: Wed, 2 Sep 1992 15:03:58 GMT
Lines: 77

In article <[e-mail]> [e-mail] (Kenneth Sloan) writes:
>In article <[e-mail]> [e-mail] (Charles Blair) writes:
>>[e-mail] (Stu Zweben) writes:
>>
>>>This is one of the main reasons that the Computing Sciences Accreditation
>>>Board was formed by ACM and IEEE-CS in the mid-80s.  Programs that are
>>>accredited through CSAB must require 2/5 of a year of science (four courses,
>>>including the equivalent of a two-semester sequence in a lab science for
>>>science majors, and ...
>>
>>   I do not see why somebody intending to programming work, even in a real
>>world setting, needs two semesters of lab science.
>I do not see why somebody intending to do programming work, even in a
>real world setting, needs a college degree at all.
>
>[of course, I also do not see why any self-respecting degree program
>needs the trappings and hassle of dealing with the CSAB]
>
I take offense at the above comment (not personally, but encompassing software
development in general). I do embedded work, and I use my operating systems theory
courses as well as digital logic and elementary circuits courses every day.

I admit my job is not typical. If it were, I'd choose some other line of work
It is some people's perception that software is this constant re-application
of old theories, and yes, sometimes it is. In other cases, especially embedded
applications, efficiency is the key. You can't run UNIX under the hood of your car,
in your microwave, VCR, camcorder, CD player, TV, or cellular phone. It's just not
efficient, and it probably never will be.
The key here is applying what you learn in theory to a real-world application,
in a cost reduced way. For application programmers, cost reduction involves less
labor (programming) to write that new application to get to the market faster
(high-level languages, C++ and object-oriented are helping in this front).
For embedded programmers, it means the use of high level languages!
With this comes faster maintenance, faster development time, and beleive it or
not, lower product development cost over the lifetime of the product.

As technology improves (thanks to the colleagues who read this newsgroup) and
more computational power is put in a smaller space, the end result is more
innovative features, and an easier to use end-product. Another by-product
is shorter development cycles (we can give you improvements at a faster pace than
ever before).

Where is this leading?? The advancement of computer architecture and it's resultant
products (RAM, EPROM, communication devices, etc) is at the highest rate it's ever
been. Processor architectures are coming out faster than they can be designed into
systems that are fully fleshed out with software. Software, by itself, becomes more
complex as systems get more powerful. From first inception to delivery Sys5R4 took
more than 2 processor development lifecycles. Software, obviously, has not kept pace.

I hate making predictions, but...
In the next 5 years, the development cost of SGI's next workstation will cost more
for software than hardware.

In summary, software curricula at major universities must keep very close tabs
on technology in industry to improve it's time to market. Software will sell
hardware, and the major processor vendors must have a lot of software support
to sell it's systems. If they don't, the ones who do, with faster development
cycles for their customers will be the "winners".

For "production" software (applications for business, engineers, etc) this means
a proficiency in X-windows, UNIX, NFS, etc. And by definition C and C++ (with an
object-oriented background). You can't do this without a college education.

For "embedded" software, this means operating systems theory, C and C++, some EE
(digital logic, elementarty circuits, etc). You can't do this without a college
education either..

The days of the guy in the corner hacking Cobol are slowly disappearing..
They are following the slow reduction in the use of the mainframe concept of computing..

-Dan

--
=> Dan DeClerck                      |  EMAIL: [e-mail]   <=
=> Motorola Intl Subscriber Group    |                                    <=
=>                                   |  Phone: [phone]             <=
----------------------------------------------------------------------------