Re: Bugs at my web site

From: glen herrmannsfeldt (gah_at_ugcs.caltech.edu)
Date: 01/20/04

  • Next message: Athanasios Migdalas: "Re: reading exact contents of a line" 
    Date: Tue, 20 Jan 2004 18:22:58 GMT

    Charles Russell wrote:

    > "*** Hendrickson" wrote

    >>What you are refering to is called "sequence association" and it's
    >>still a first class citizen in F2003.

    > Was it actually in the written standard for f77 or earlier?
    > Or was it just a part of fortran tradition?

    Fortran 66 has:

    "7.2.1.1.1 Array Element Successor Function and Value of a Subscript.
    For a given dimensionality, subscript declarator, and subscript, the
    value of a subscript pointing to an array element and the maximum
    value a subscript might attain are indicated in Table 2. A subscript
    expression must be greater than zero.

    The value of the array element successor function is obtained by adding
    one to the entry in the subscript value column. Any array element whose
    subscript has this value is the successor to the original element. The
    last element of the array is the one whose subscript value is the
    maximum subscript value and has no successor element.

                  Table 2 Value of a Subscript

    Dimensionality Subscript Subscript Subscript Maximum
                     Declarator Value Subs. Value

            1 (A) (a) a A
            2 (A,B) (a,b) a+A*(b-1) A*B
            3 (A,B,C) (a,b,c) a+A*(b-1)+ A*B*C
                                           A*B*(c-1)

    a,b,c are subscript expressions
    A,B,C are dimensions."

    This seems to allow you to reference an array element with any
    combination of subscripts that generate the same "Subscript
    Value", though all subscript expressions must be greater
    than zero.

    Part of 8.4.2 "If a dummy argument of an external function is an array
    name, the corresponding actual argument must be an array name or array
    element name."

    also in 8.4.2 "The actual arguments, which constitute the argument
    list must agrees in order, number, and type with the corresponding
    dummy arguments in the defining program."

    Note that it does not mention that dimensionality must agree.

    That specifically allows an array element to be passed to a
    subprogram expecting an array. Then, in 10.1.3...

    "If the actual argument is an array element name, the length of the
    dummy argument array must be less than or equal to the length of the
    actual argument array plus one minus the value of the subscript of
    the array element."

    This seems to be discussed in terms of dimensionality of one,
    but also in subscript value. That, and the ones described
    above, seem to imply that arrays are stored in the order
    described by the array element successor function, and can
    be referenced that way in subprograms. Note, though, that
    in 7.2.1.1 Array Declarator...

    "No array element name may contain a subscript that, during
    execution of the executable program, assumes a value less than
    one or larger than the maximum length specified in the array
    declarator."

    This would seem to exclude the popular (1) declarator, though
    it does not exclude a dummy array declarator with a very large
    number in it. For implementations that do bounds checking, this
    could be important.

    For EQUIVALENCE statements it is specifically allowed to refer
    to an array with one subscript, and use the array element successor
    function to determine which element is being described.

    -- glen

  • Next message: Athanasios Migdalas: "Re: reading exact contents of a line"
    • Quantcast