From: brunobg AT geocities DOT com (Bruno Barberi Gnecco)
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Trouble with many mallocs and reallocs
Date: 10 Aug 1998 15:24:02 -0500
Organization: Newscene Public Access Usenet News Service (http://www.newscene.com/)
Lines: 50
Message-ID: <35cf519a.8496651@news3.newscene.com>
References: <Pine DOT SUN DOT 3 DOT 91 DOT 980810093926 DOT 1429A-100000 AT is>
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

On Mon, 10 Aug 1998 09:39:50 +0300 (IDT), Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
wrote:

>On 9 Aug 1998, Bruno Barberi Gnecco wrote:
>> Debugging it, I find that the error is exactly at the line that
>> allocates memory to a new line (it allocates 1 byte). I'm running
>> the Windows 95 DPMI server. What may be the problem?
>You need to supply more information to get meaningful replies.  At the
>very least, explain what is that ``error exactly at the line that
>allocates memory to a new line''.  Please describe what is the problem
>that you see under a debugger.

The node is:

struct node {
  struct node *forward, *back;
  unsigned char *data;
  };
typedef struct node NODE;

and the trouble is at

int insert_node( unsigned char *p, int d) {
  node *nn_p;
  unsigned char *nd_p;

  nn_p = ( NODE * ) malloc ( sizeof ( NODE ));
  nd_p = ( unsigned char * ) malloc ( strlen(p) + 1 );

p value was '\0', and I still can't see any special reason to it happen. I tried
to type a line over 1kb in length, and it worked perfectly, but 20 lines of 3
characters didn't. The reallocation is OK. I debugged it step by step twice, and
these particular times the trouble happened at the nn_p line, returning a SIGABRT,
and in another point, also returning a SIGABRT.  I realy don't believe is lack of
RAM, because I used less than 1kb in variables, and the .EXE is 195kb length
(unoptimized and with debugging info).

>> Is there a better way to allocate memory?
>Yes.  The usual method is to allocate memory for the buffer, and then
>maintain pointers into that buffer which point to the beginning of
>each line.  There are other methods as well.  For example, an editor
>could be totally oblivious to the line-oriented structure of text, and
>instead treat it as a stream of bytes.

I'm doing the first method; I first allocate a temporary buffer, then, if
everything is alright, pass the address to the node.data.ú

"There's never enough time to do all the nothing you want" Bill Watterson 
Bruno Barberi Gnecco <brunobg AT geocities DOT com> ICQ #1383173 - PGP 5.0i user 
My other OS is Linux -=- http://graphx.home.ml.org -=- Electric Eng, Poli