Author: Robert Hyatt
Date: 12:17:22 01/10/03
Go up one level in this thread
On January 09, 2003 at 19:56:44, Dieter Buerssner wrote:
>On January 08, 2003 at 18:28:56, Robert Hyatt wrote:
>
>>On January 08, 2003 at 16:01:34, Dieter Buerssner wrote:
>>
>>>On January 08, 2003 at 15:45:27, Robert Hyatt wrote:
>>>
>>>>On January 07, 2003 at 09:16:21, Uri Blass wrote:
>>>>
>>>>>2)I get a lot of remarks for movei and the most popular remark is
>>>>>
>>>>>external definition with no prior declaration
>>>>>
>>>>>Do I need to declare function before I use it.
>>>>
>>>>Yes you must.
>>>
>>>Hmmm - I don't see this. Can you cite a part of the C Standard, that shows, that
>>>you really "must"?
>>
>>Yes.
>>
>>The part that says "if a function is not declared before it is used, the
>>assumptions are
>>..."
>
>To me it seems, you are misinterpreting it here. When a function is defined, it
>is also imlicetly declared. Note, that the warning in Uri's case was not at the
>place, where the function was used, but at the place, where the function was
>defined.
I understand that. But that assumes that (a) the function is defined in the
_same_
file before it is used and (b) the function is defined _before_ it is used. I
can show
you a program where this is impossible to do, which means that explicit
declarations
for the functions have to either appear directly in the procedures that call
them, or
more commonly, in a header file everything includes...
>
>int foo(void)
>{
> return 1;
>}
>
>int main (void)
>{
> return foo();
>}
>
>Is perfectly legal C. As Heiner pointed out, it can be considered bad style. If
>one really need to have foo with external linkage, a prototype will be needed
>somewhere. Otherwise, static int foo() would be more appropriate.
>
>>:)
>>
>>By the same token, You should always use signed or unsigned when declaring chars
>>since the standard leaves that an open issue as well...
>
>I understand your point here, but I don't agree in this generality. You need
>signed or unsigned, when you use char as arithmetic types, and the range 0-127
>is not enough for this purpose. For a single flag (0 or 1) for example, you
>don't need signed or unsigned. When using them for strings, signed or unsigned
>would probably yield in many problems. All the str... functions want pointer to
>char, and not pointer to signed char or pointer to unsigned char.
>
>>>>>5)operands are evaluated in unspecified order
>>>>>
>>>>>It is not my fault that the intel compiler does not know that the order of
>>>>>adding is not important because it does not know that calculating one expression
>>>>>does not change the second expression.
>>>>>
>>>>>Uri
>>>>
>>>>The order _is_ specific. That usually means you are mixing operators of
>>>>difference
> ^^^^^^^^^^
>>>>precedence without using () to specify your chosen order of evaluation. Just
>>>>add
>>>>the parens to avoid some _ugly_ errors...
>>>
>>>The order, in which different terms are evaluated in aritmetic expressions is
>>>typically _not_ specific in C. Mixing operators of _different_ precedence
>>>without () is no problem at all. Normally, one would not want to write
>>>
>>>w = ((x*y) + z);
>>
>>
>>Sure... but what about the bitboard stuff that includes
>>
>>a>>n+1
>>
>>That can wreck a program silently and hideously...
>>
>>Depending on what you mean...
>
>Sure, but this was not the point here. The point was, that even in
>
>w = (x() * y()) + z());
>
>there is no guarantee, that x() or y() are called before z(). How the function
>returns values are put together, is (with or without parentheses) well defined.
>
>Regards,
>Dieter
This page took 0.02 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.