Re: Large University Software Projects

Re: Large University Software Projects

[cis.ohio-state.edu!magnus.acs.ohio-state.edu!zaphod.mps.ohio-state.edu!mi]

srctran@world.std.com (Gregory Aharonian) writes:
> The latter part of the above statement is incorrect.  There are many large,
> long term software engineering projects going on at universities that student
s
> participate in, both as thesis projects, and as class subjects, large in the
> sense of systems dealing with hundreds of thousands of lines of code.
...
Does the average undergrad computer science student work on these large
projects?
 
When you use the phrase "thesis projects" I would assume, perhaps wrongly,
that these projects use only honors undergrads or graduate students.  In
my humble opinion, all students would benefit from the experience of
working on "long term software engineering projects."  I believe that
every person graduating with a degree in Computer Science should have
worked on at least one large program (100,000 SLOCs or more).

By the way, could you state the language used for these large projects?
 
Robert Parkhill 
Paramax, A Unisys company

Re: Large University Software Projects

[Gregory Aharonian]

>Does the average undergrad computer science student work on these large
>projects?  ...plus a similar comment
Based on statistically insignificant data, I would say not.  Mostly likely
not introductory CoSci stduents (freshmen,sophmores), but maybe some juniors
and more seniors. It depends on the university and the size of the department.
Also a fair number of undergrads have summer jobs in the departments, where
they are exposed to these systems.

>By the way, could you state the language used for these large projects?

For most science/applied engineering applications, the language is Fortran,
though there has been some transition to C/C++ in recent years, usually by
the engineering departments.
For most computer science/electrical engineering software, C has been
strong through the 1980's, with a gradual transition to C++.  At the large
schools (MIT,Berkeley,Stanford,etc) where some of the big projects go on
(X-Windows,UNIX,Database stuff,Electronic Design) it is pretty much all C/C++.

Fortran and C make up at least 80% of the large applications.  Some of the
other language include Basic,Pascal,1-2-3,Prolog,Lisp,Dynamo, and SLAM.
The only place you see Ada in any significant volume are the DoD gradaute
schools (Monterey,WrightPat), and some DoD-funded university projects.
Partly this is due to compiler accessibility, and despite the dreams of the
Ada-9X teams, it is too late for Ada to make major inroads in the colleges.

The impressions made by the Mosemann studies with regards to the acceptance
of programming languages for usage and instruction in the universities are
totally misleading, and are based on a very poor survey of language usage.
Had they actually visited some universities, and traversed over USENET, they
would have been able to collect enough data to arrive at conclusions different
from those in the studies.

The continual apathy by the DoD community to collecting real data about
issues such as language use, software reuse, and software technology transfer
will continue to lead to less than optimal software engineering decisions
made by the DoD.  As read some of these reports, I get the impression that
people are reporting what they think wants to be heard.

I mean, all you have to do is walk into your local university library,
and flip through a few thousand theses going back a few years.  Ada is
statistically insignificant as a language of choice.  In fact, if anyone
is ever in the Boston area, I would be glad to show them around some of
these libraries, and let you count for yourself.

Ada has been marketed poorly by the DoD to these outside communities like
the academic world.  You might argue that the DoD shouldn't be in the role
of marketing, but with the policies they set and shaky studies they fund,
they have assumed that role.

In most arenas where choice of language is not forced, Ada is losing out
(for whatever reasons), and in the long run this will drive up the costs of
Ada development by anyone, as the markets devote their resources to languages
like C++.

The DoD can either take Ada more seriously, and fund studies that show why
Ada is not succeeding economically, or the DoD can allow more than one
language to be official.  In either case, the microeconomic treatment and
studies of Ada and defense software engineering have been more self-serving
tham self-consistent.

Greg Aharonian
Source Translation & Optimization