NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/comp/lang/c/19102

Internet Message Format | 1992-12-31 | 1.8 KB

Path: sparky!uunet!spool.mu.edu!agate!dog.ee.lbl.gov!horse.ee.lbl.gov!torek
From: [e-mail] (Chris Torek)
Newsgroups: comp.lang.c
Subject: Re: Moving from Pascal to C, Help please!!!!!!
Date: 1 Jan 1993 01:59:41 GMT
Organization: Lawrence Berkeley Laboratory, Berkeley CA
Lines: 35
Message-ID: <[e-mail]>
References: <[e-mail]> <[e-mail]>
NNTP-Posting-Host: 128.3.112.15

In article <[e-mail]> [e-mail]
(Brian Katzung) writes:
>#define    ARRAYLOWER    6
>#define    ARRAYUPPER    26
>int    _array[ARRAYUPPER - ARRAYLOWER + 1];
>#define    array    (&_array[-ARRAYLOWER])

As the FAQ warns (question 2.11), this technique is not strictly
kosher.  The effect of computing an out-of-bounds address is
undefined.  Most implementations will `do what you mean' here, but
there are no guarantees.

Note that the trick is fully portable if and only if the zero point
is contained within the array (i.e., ARRAYLOWER <= 0 < ARRAYUPPER),
because then &_array[-ARRAYLOWER] is within the array.  E.g., if
the lower bound is -3 and the upper bound is 1, then we have

    int    _array[1 - (-3) + 1];

or

    int    _array[5];

and -(-3) is 3, so we compute &_array[3], which is legal (3 < 5).
(In fact it works when ARRAYUPPER is 0 too, because given an array
N of T, we can always compute &array[N]; we are just not allowed to
try to *use* array[N].)

Incidentally, the name `_array' is reserved to the implementation when
external, so to make this right we also have to use some other name, or
make `_array' static or automatic.  (Technically, _array is safe if
and only if we give it `internal linkage' or `no linkage'.)
--
In-Real-Life: Chris Torek, Lawrence Berkeley Lab CSE/EE ([phone])
Berkeley, CA        Domain:    [e-mail]