NetNews Usenet Archive 1992 #30/NN_1992_30.iso/spool/comp/lang/c/18310

Text File | 1992-12-14 | 2.6 KB | 66 lines

Nntp-Posting-Host: solva.ifi.uio.no
Newsgroups: comp.lang.c
Path: sparky!uunet!mcsun!sunic!aun.uninett.no!nuug!ifi.uio.no!nntp.ifi.uio.no!jar
From: [e-mail] (Jo Are Rosland)
Subject: Re: Question to test general C knowledge
In-Reply-To: [e-mail]'s message of Sun, 13 Dec 1992 05:14:42 GMT
Message-ID: <[e-mail]>
X-Md4-Signature: 26f0d8e13c00a2845a2e57509e2389f6
Sender: [e-mail] (Jo Are Rosland)
Organization: Dept. of Informatics, University of Oslo, Norway
References: <[e-mail]> <[e-mail]>
    <[e-mail]>
    <[e-mail]>
    <[e-mail]>
Date: Sun, 13 Dec 1992 22:54:54 GMT
Lines: 47
Originator: [e-mail]

In article <[e-mail]> [e-mail] (Christopher R Volpe) writes:
   |>
   |> Based on a sequence like
   |>
   |>         ! compute needed values
   |>         addl3   _i,$1,r0        ! compute value of ++i
   |>         ! sequence point; do accumulated side-effects
   |>         movl    r0,_i           ! store value of ++i into i
   |>         incl    _i              ! increment i
   |>
   |> Thus we see the wisdom of ANSI's rule :-)

   Umm. I question whether it would be legal for the compiler to emit
   such code. The reason is that the value of the "++i" expression is
   "the value of i after the assignment". The first assembly statement
   computes this value as "3". However, the "value of i after the assignment"
   is determined by the result of the "incl _i" statement, and that
   value is 4, which contradicts the definition of the value of the "++i"
   expression.

An expression in C has a return value.  Some expressions also have
side effects.  The _return_value_ of "++i" is 3 (given 2 as the
original value), while the _side_effect_ is to increment "i".  Trouble
is, the compiler doesn't have to compute the return value and execute
the side effect at the same time.

The following is the likely way the compiler will operate:

- The return value of "++i" as well as the address of "i" will have to
  be computed before the "=" is performed.

- The side effect of incrementing "i" can be performed before, during
  or after these computations.

- The compiler can perform the incrementing of "i" by assigning it the
  return value of "++i" at some point after it's computed, or by simply
  incrementing it.

But, due to the "undefinedness" of the construct, none of these are
guarantees.  The compiler could opt to do something entirely
unexpected, like bringing about world peace.

Hope this helps,

--
Jo Are Rosland
[e-mail]