NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/lang/c-part1/6186

Internet Message Format | 1996-08-05 | 3.5 KB

Path: mail2news.demon.co.uk!genesis.demon.co.uk
From: Lawrence Kirby <[e-mail]>
Newsgroups: comp.lang.c
Subject: Re: Easiest way to center a string?
Date: Fri, 23 Feb 96 00:02:17 GMT
Organization: none
Distribution: world
Message-ID: <[e-mail]>
References: <4fobka$[e-mail]> <[e-mail]> <[e-mail]>
Reply-To: [e-mail]
X-NNTP-Posting-Host: genesis.demon.co.uk
X-Newsreader: Demon Internet Simple News v1.27
X-Mail2News-Path: genesis.demon.co.uk

In article <[e-mail]>
           [e-mail] "The Amorphous Mass" writes:

>On 21 Feb 1996, Tanmoy Bhattacharya wrote:
>
>> The Amorphous Mass <[e-mail]> writes:
>> <snip>
>>    > In article Amorphous Mass <[e-mail]> writes:
>>    > <snip>
>>    >      That's equivalent to strlen(String + -LineLen), which is in turn
>>    >    equivalent to strlen(String[-LineLen]), and the expression
> String[-80] is
>>
>>                              ^ this is now a char, not a char*.
>
>  Certainly is.  My apologies for slipping up here.  For some
>unaccountable reason I was thinking a[i] was equivalent to (a + i) and not
>*(a + i).
>
>>    >      If you meant (strlen(String) - LineLen) / 2 then you will get a
>>    >    negative length specifier for any string with fewer than LineLen
>>    >    characters.  If you meant (LineLen - strlen(String) / 2, then your
>>    > strlen returns size_t which is guaranteed unsigned. (I am assuming
>>    > LineLen is not long.)
>
>>      Right, but strlen(String) would be 12, minus LineLen which is 80, which
>>    is -68, divided by two is -34, which is the number passed as a width
>
>> but if size_t is unsigned int or unsigned long, and LineLen is int,
>> the result is not -68 because an unsigned quantity cannot be -68. (My
>> parenthetical remark was wrong: I had assumed that size_t was at least
>> as long as unsigned int and LineLen is not long).
>
>  I guess I still need to have this clarified (my knowledge of type
>promotion et al is admittedly shaky).  If you subtract an unsigned int
>from an int, is the result necessarily unsigned int?

Yes, this follows from the rules for the 'usual arithmetic conversions'.
These ensure 2 operands are promoted to a common type. I won't list them
out in full right now but if one operand is a signed long and the other
an unsigned long the result is unsigned long. Similarly for signed and
unsigned int. In that respect I believe the ANSI rules match the old
K&R rules.

>Is that true
>generally for integral types?

All chars and shorts promote to int except where an int can't represent all
of the values of an unsigned char or unsigned short. in that case it
promotes to unsigned int. The other 'usual arithmetic conversion' rules are
then applied to these promoted types.

> IOW, are you saying that strlen("hello,
>world") - 80 will evaluate to UINT_MAX - 68 (or something like that)?

I made a mistake. -68U evaluates to (UINT_MAX+1)-68, not UINT_MAX-68.

If size_t corresponds to unsigned int or unsigned long that is correct.
If it corresponds to unsigned short or unsigned char and int is wider
then the result will have type int and value -68. If you want to ensure that
value you should write (int)sizeof("hello, world") - 80

--
-----------------------------------------
Lawrence Kirby | [e-mail]
Wilts, England | [e-mail]
-----------------------------------------