eberard@ajpo.sei.cmu.edu (Edward Berard) (07/20/88)
The July 11, 1988 issue of Federal Computer Week has an editorial
written by Fred Reed ("Ada: A View From the Field", page 19) which
supposedly presents both the pros and cons of Ada. The following
quotes are representative:
"What follows is one man's opinion, but it is an informed opinion --
he programs in Ada as well as C and other -- so I thought it was worth
passing along. We'll call him Bob because, Ada being a fairly sacred
cow, he didn't want to be named."
"... Ada has been called everybody's favorite 17 programming
languages."
"Fewer programming errors. Among other features, Bob explained, Ada
has what are called strong typing and range checking. Strong typing
means that the programmer is forced to tell the computer the nature of
each piece of information he uses. This allows the machine to be sure
incompatible kinds of information are not mixed."
"Ada, said Bob, is well suited to being understood and maintained by
many programmers who will work on a large program over its lifetime.
This is why it saves money, he said."
"However, he noted, these strength also result in weaknesses. For
example, the programmer has to write extra code to do the strong
typing and range checking, and this, among other things, makes Ada
code slow to write."
"Further, because the program constantly has to check ranges and data
types as it runs, Ada produces much slower programs than some other
languages.
"An Ada program typically runs a tenth as fast as the same program in
C," said Bob. "I think we are seeing the bureaucratic attitude at its
ultimate. Ada is a prudent approa
ch to software. It is defintely
suboptimal in performance -- slow, bulky, verbose and slow to write.
But it is reliable and easy to work with after it is written. It is
safe. These are qualities prized by bureaucrats. DoD chose reliability
over performance."
-- Ed Berard
(301) 695-6960leake@cme-durer.ARPA (Stephe Leake) (07/21/88)
I have done a study (funded by NASA for the Space Station) comparing
Ada and C on Suns and microVAXes. The application was robot
kinematics, and (briefly), the results were:
Ada was easier to code and debug. The Ada development environment (DEC
LSE, VMS) was _much_ nicer than the C environment (DEC C - not a very
friendly compiler). Also, I was able to code more abstractly in Ada,
making the code match the problem statement, which eases debugging.
Since this work, I have done other work in Ada and C, and I find I am
_much_ more productive in Ada (although some of that productivity
comes from VMS vs UNIX).
DEC Ada 1.4 was _Faster_ (by about 25%) than DEC C 2.3 (Hip Hip Hooray!!!)
(honesty forces me to admit that some of the speed gain came
from using single precision - but how do you get C to use single precision?)
Verdix Ada 5.5 (on the Sun) was slower (by about 10%) than Sun C 3.2
( good enough for me)
I am interested in publishing this work - can anyone recommend an
appropriate journal?
Anyone else have similar data? Its time us Ada users spoke up.
--
NAME: Stephe Leake TELE: (301) 975-3431
USMAIL: National Bureau of Standards ARPA: leake@cme-durer.arpa
Rm. B-124, Bldg. 220 UUCP: uunet!cme-durer!leake
Gaithersburg, MD 20899smithrd@inteloa.intel.com (Randy D. Smith) (07/26/88)
In article <521@marvin.cme-durer.ARPA> leake@cme-durer.ARPA (Stephe Leake) writes: >I have done a study (funded by NASA for the Space Station) comparing >Ada and C on Suns and microVAXes. >... >DEC Ada 1.4 was _Faster_ (by about 25%) than DEC C 2.3 (Hip Hip Hooray!!!) ^^^ ^^^ ^^^^^^^^^ > (honesty forces me to admit that some of the speed gain came ??????? >from using single precision - but how do you get C to use single precision?) ... >I am interested in publishing this work - can anyone recommend an >appropriate journal? > >Anyone else have similar data? Its time us Ada users spoke up. >-- >NAME: Stephe Leake TELE: (301) 975-3431 >USMAIL: National Bureau of Standards ARPA: leake@cme-durer.arpa Might I suggest "The Journal of Skewed Results"??!? You might call it, "I Found What I Was Looking For"!!?! :-) Seriously now, I would like to believe you, but the fact that you had such an obvious bias would make it difficult for me to accept your study as objective. Perhaps you'd like to provide us a little more detail on your method for determining which code to use for your comparison, and which to throw out. Hmmmm, a quick check of Webster's confirms that this appears to meet the definition of propaganda. Oh well.... -- Randy D. Smith BiiN (tm) An Information Systems Company ...uunet!tektronix!ogcvax!inteloa!smithrd (503) 696-4660