I don't go back to the beginning of FORTRAN. My comment was not that FORTRAN
was badly named when it was among the first to do such things. I am saying
that in retrospect, almost any language can do a basic subset of arithmetic
operations. And there is nothing in principle that necessarily stops any
modern language code from being optimized by translators to be even more
rapid than the original versions of FORTRAN. If anything, I can well imagine
algorithms using parallel architectures from performing some operations way
faster. True, many languages tend to add overhead but that is not
necessarily required.
As some have said here, many things these days are fast enough not to need
ultimate optimization.
Having said that, there is no reason why code used over and over should not
be optimized. Functions in higher-level languages can be written using the
language and then can be replaced if there is a great enough improvement and
they do not need some of the interactive features the language might offer.
Someone mentioned that in principle all data types can be stored in a
string. Python does have a concept of converting many, but not all data
structures into a byte form that can be written to files and restored or
sent to other processes including on other machines. Of course entire
programs written in text can be shipped this way to some extent. But when
your data is a long list of real numbers that you want to find the standard
deviation of, then converting it into a compact C/C++ array (or something
similar in Fortran) and calling a function in that language that works fast
on that, may be a good way to go. Why interpret a step at a time when one
pass generates the data that can be processed in tighter loops and a result
returned?
-----Original Message-----
From: Python-list <python-list-bounces+avigross=verizon....@python.org> On
Behalf Of Dennis Lee Bieber
Sent: Friday, January 4, 2019 1:17 PM
To: python-list@python.org
Subject: Re: the python name
On Fri, 4 Jan 2019 11:34:24 -0500, "Avi Gross" <avigr...@verizon.net>
declaimed the following:
>
>Although I used FORTRAN ages ago and it still seems to be in active use, I
am not clear on why the name FORMULA TRANSLATOR was chosen. I do agree it
does sound more like a computer language based on both the sound and feel of
FORTRAN as well as the expanded version.
>
>It seems to have been designed as a mathematical extension of sorts that
allowed you to evaluate a mathematical formula efficiently. I mean things
like quadratic equations. But there is overlap with what other languages
like COBOL or BASIC did at the time.
>
FORTRAN predates BASIC by a decade.
COBOL was never meant to be an in-depth number cruncher language
(original data type is packed BCD). Writing a quadratic equation in it
probably takes two or three pages (as I recall, the COMPUTE verb was a later
addition, so everything would have been long sentences:
DIVIDE C BY B GIVING TMP1. MULTIPLY TMP1 BY 2. SUBTRACT TMP1 FROM A GIVING
RESULT1. ADD TMP1 TO A GIVING RESULT2.
vs
COMPUTE TMP1 = (C / B) * 2. COMPUTE RESULT1 = A - TMP1. COMPUTE RESULT2 = A
+ TMP1. )
>What gets me is the vagueness of the words looked at by ME today. Any
modern computing language can do what standard FORTRAN does, albeit perhaps
more slowly as I know some languages do some of their math using libraries
from FORTRAN. But do we use the word TRANSLATOR quite that way much anymore?
Heck, do we use FORMULA in the same way?
Meanings change... "COMPUTER" means "one who computes" -- those poor
overworked engineers with slide-rules creating ballistic tables for
battleships.
"Translator" still applies -- in the sense of taking one language
(source) and producing the equivalent meaning in another language
(assembler, and then translating that to pure machine binary). "Formula"
has likely been superceded by "algorithm" (cf ALGOL)
>
>My most recent use of formula has been in the R language where there is a
distinct object type called a formula that can be used to specify models
when doing things like a regression on data. I am more likely to call the
other kind using words like "equation". Python has an add-on that does
symbolic manipulation. Did FORTRAN have any of these enhanced objects back
when created, or even now?
No body had symbolic manipulation when FORTRAN was created. At the
time, the goal was to produce a higher order language that the scientists
could write without having to /know/ the computer assembly/machine code,
with the hope that it could be portable at the source level. Users were
charged by the minute for CPU time, and operating systems were lucky to be
able to handle more than one program in parallel. Batch systems were the
norm -- where a program would run until it either finished, or it ran out of
CPU time (as specified on a LIMIT in the job control deck). One would turn
in a deck of cards to be spooled in the job queue, and come back some hours
later to get the printout from the job.
--
Wulfraed Dennis Lee Bieber AF6VN
wlfr...@ix.netcom.com HTTP://wlfraed.home.netcom.com/
--
https://mail.python.org/mailman/listinfo/python-list
--
https://mail.python.org/mailman/listinfo/python-list
