Efffect of Language Level of Effort -- Feb 4, 1997 -- TCACMSIGAda

Efffect of Language Level of Effort -- Feb 4, 1997 -- TCACMSIGAda

[Paul D. Stachour]

Twin Cities SIGAda presents:

	      Effect of Language Level on Project Effort
	 Paul Stachour, Programming Alternatives of Minnesota

          7-9 PM, Tuesday February 4th, 1997
                    Oxboro Library
          8801 Portland Avenue, Bloomington, Minnesota.


In the early 1970's, there was significant debate on the relative
merits of assembly vs systems-implementation languages.  Many people
held that the necessary efficiency and control could only be gained
through coding "Close to the Machine".  Eventually we learned that
higher global efficiency was possible only with systems-implementation 
languages.  In the late 1980's, and continuing today, we still have 
people who insist that only with low-level languages such as "C" can
we have efficiency and control.

Barry Boehm, with his COst COnstructive MOdel, showed us that the 
primary driving factor behind project costs was the effort needed
to design, write, and test a source-line-of-code (SLOC).  He also,
through regression analysis, showed what other cost factors highly
influence the amount of effort required on a project.

From data gathered over many projects in the 80's and 90's, the
Software Productivity Institute has provided a set of numbers,
the "language level", that relates how many source-lines-of-code are
necessary to implement a given amount of function (a function-point).

Tucker Taft, of Intermetrics, has written a paper showing why Ada
programs can be compiled to more efficient code than C programs.
The specifics are different, but the reasoning (what is visable to
the compiler, and how the compiler can take advatage of it) is the
same as that for the assembler versus systems-implementation languages.

Paul Stachour will relate all 4 of the topics above, and then 
demonstrate, though use of a COCOMO program created at USC,
how the language-level of Ada gives it such an overwhelming advantage
that even a expertly-trained "C" staff (all programmers have 5 years
or more of experience) still takes about 20% more effort than a 
totally untrained (no experience at all) Ada staff.  Paul will 
"walk us through the numbers" and plug into COCOMO any assumptions 
you would care to specify about the 4 main COCOMO factors: 

  Project, Computer, People, and Project.

     Oxboro Library may be reached by taking the Portland Avenue exit
from Interstate 494 in Bloomington, and driving about 10 blocks south.

     This meeting is open to the public.  Twin Cities SIGAda is a
special interest group of the Association for Computing Machinery
(ACM).  For more information regarding this meeting or membership
in Twin Cities SIGAda contact Chad Broshar at 871-1339, Email to
brosha19@skypoint.com, or point your WWW browser to
"http://www.winternet.com/~stachour/ada/tcacmsigada.html" or 
"http://www.winternet.com/~stachour/ada/feb97mtg.htm". 



--
Paul D. Stachour, Software Engineer and Methodologist
9532 First Avenue South                 United Health Care, MS W131
Bloomington, Minnesota 55420            5300 Olson Mem Hwy, Golden Valley, MN
(612)-884-5977  stachour@winternet.com  (612)-204-8872  pstacho@ducttape.uhc.com

Re: Efffect of Language Level of Effort -- Feb 4, 1997 -- TCACMSIGAda

[Ted Dennison]

Paul D. Stachour wrote:
> 
> Tucker Taft, of Intermetrics, has written a paper showing why Ada
> programs can be compiled to more efficient code than C programs.
> The specifics are different, but the reasoning (what is visable to
> the compiler, and how the compiler can take advatage of it) is the
> same as that for the assembler versus systems-implementation languages.

Is this paper available on the net anywhere? It sure would be a useful
reference when this issue comes up again here. (And you all know it
will).

-- 
T.E.D.          
             |  Work - mailto:dennison@escmail.orl.lmco.com  |
             |  Home - mailto:dennison@iag.net               |
             |  URL  - http://www.iag.net/~dennison          |

Re: Efffect of Language Level of Effort -- Feb 4, 1997 -- TCACMSIGAda

[Ken Garlington]

Paul D. Stachour wrote:
> 
> Tucker Taft, of Intermetrics, has written a paper showing why Ada
> programs can be compiled to more efficient code than C programs.
> The specifics are different, but the reasoning (what is visable to
> the compiler, and how the compiler can take advatage of it) is the
> same as that for the assembler versus systems-implementation languages.

If anyone has a copy of this paper, I'd be interested.

--
LMTAS - The Fighter Enterprise - "Our Brand Means Quality"
For job listings, other info: http://www.lmtas.com or
http://www.lmco.com