Path: utzoo!attcan!uunet!lll-winken!ames!haven!umbc3!mbph!hybl
From: hybl@mbph.UUCP (Albert Hybl  Dept of Biophysics  SM)
Newsgroups: comp.lang.fortran
Subject: An exercise in futility
Keywords: The RIGHT number
Message-ID: <588@mbph.UUCP>
Date: 18 Jan 89 16:16:07 GMT
Organization: University of Maryland, School of Medicine, Baltimore, MD 21201
Lines: 73


In message <454@loligo.cc.fsu.edu> mccalpin@loligo.cc.fsu.edu
(John McCalpin: Supercomputer Computations Research Institute)
writes:
>In article <587@mbph.UUCP> hybl@mbph.UUCP (Albert Hybl 
>Dept of Biophysics  SM) writes: Consider the following program:
>>*  Increase the value of K until the program fails.
>>      PARAMETER (K=24)
>>      CHARACTER AND(K)*1,DNA*(K)
>>      EQUIVALENCE (DNA,AND)
>>      DNA(K-23:K) = 'An exercise in futility!'
>>      PRINT *,K,(AND(J+K-24),J=1,24)
>>      END
>>What does the ANSI X3.9-1978 standard say are the maximum allowable
>>sizes for the character array AND(K)*1 and the character variable
>>DNA*(K)?
>
>Of course *you* are waiting in the wings with the *right* answer to
>this problem.  What number do you suggest?

I would like the maximum allowable sizes for the character
array AND(K)*1 and the character variable DNA*(K) to be the
same!!!....  If they were the same, then PARAMETER(K=nnnn)
could be used to adapt a program to fit whatever available 
core the target machine may have.  I think that the existence
of the PARAMETER statement and the vagueness of the ANSI
X3.9-1978 standard were meant to allow for processor size
differences and not to give implementors a vehicle to demolish the
standard.

>...                                   It sounds like you want a number
>of several millions so that you can fit your genome sequences in.  That
>is nice, except that it means that in *your* world, no standard IBM PC
>could ever have a valid Fortran language system, because there ain't
>enough core....

      CHARACTER AND(30000)*1     ! Is allowed on an IBM PC/RM fortran
      CHARACTER DNA*256          ! Is NOT--the compilation fails!

Believe me, I want a portable fortran for the IBM PC, but when
the implementor limits the string length of a character variable
to the range 1-255, he has entered a possible winner in the
FORTRAN demolition derby!  I was pleased to see that the AT&T
XLA+ fortran compiler do what I think the standard specifies;
that is, the EQUIVALENCE (DNA,AND) statement works over the whole
range of allowable AND(K)*1.  As to wanting several million
bytes, I'd use them if they were available; but--please give me the
dynamic memory allocation routines like malloc, calloc, realloc,
and free from the C language.  I see no need to over utilize
core, when I can process only a portion of the character
string/array at a time.

>...            I, for one, would like any standard-conforming Fortran
>system to include a document page listing the implementation-dependent
>limitations, e.g.:
>	maximum size of character string (e.g. char*K string)
 	maximum size of character array  (e.g. CHAR*1 ARRAY(K))
>	maximum integer value
>	maximum and minimum floating-point values
>	maximum array size (e.g. 64kB on old IBM PC compilers)
        maximum bytes per WRITE N, LIST
>	argument range limits for ALL intrinsic math functions
>	etc....

Hmmm, that sounds like an exercise in futility.

----------------------------------------------------------------------
Albert Hybl, PhD.              Office UUCP: uunet!mimsy!mbph!hybl
Department of Biophysics       Home   UUCP: uunet!mimsy!mbph!hybl!ah
University of Maryland                CoSy: ahybl
School of Medicine
Baltimore, MD  21201                 Phone: (301) 328-7940 (Office)
----------------------------------------------------------------------