From: "Cabra" <la DOT cabraLOBAILAO AT usa DOT net>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: yet more questions about pointers
Date: Tue, 8 Sep 1998 16:32:12 +0200
Organization: Que me quiten LOBAILAO
Lines: 30
Message-ID: <6t7bg7$pih$1@diana.bcn.ibernet.es>
References: <01bdc7e8$c1ff0e20$4ac3b8cd AT scully> <35D5B41F DOT 4534F8E4 AT unb DOT ca> <35e754f4 DOT 2457929 AT cnews DOT newsguy DOT com>
NNTP-Posting-Host: 194.75.70.157
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

wpp AT brinet DOT com escribió en mensaje <35e754f4 DOT 2457929 AT cnews DOT newsguy DOT com>...
>On Sat, 15 Aug 1998 13:15:28 -0300, Endlisnis <s257m AT unb DOT ca> wrote:
>
>>Cephaler wrote:
>>
>>> int main(void) {
>>>   char *string1=(char *)malloc(80);
>>>   char *string2;
>>>   strcpy(string1,"foobar");
>>>   string2 = string1;
>>>   strcpy(string2,"raboof");
>>>   printf("%s\n",string1);
>>>   return(0);
>>> }

[...]
>>> 3) concerning strcpy...is there any special reason why I shouldn't just use
>>> string1 = "foobar" ?
>>
>>    Not to my knowledge (except you can't 'free' that space).

In 'Cygnus' GCC, this definition is considered a 'non-const' pointer to a 'const'
bunch of bytes, i.e. you can point to it, but you can't modify this contents.

You can define 'char string1[]="foobar"' to avoid this behaviour.

DJGPP (with the aproppiate compiler options) warn this, but IIRC, 'Cygnus' not.