Re: Proposed patch: Modify glib to permit embedded NULs in GString.

[Owen Taylor <otaylor redhat com>]

I agree that since it isn't very hard to do, GString should
support embedded NULS. 

There is a question of how far to 
go with this - almost any string function could conceivably
have a variant that handles embedded \0. My opinion on this
is that we should stop here - most of the time when you have
a string with embedded \0 it is basically a data chunk.

But having this basic capability will be useful.

I've applied your patch with only a few changes:

> +  while (i) {
> +    if (*p != *q)
> +      return FALSE;
> +    p++;
> +    q++;
> +    i--;
> +  }

Indentation here needed to be fixed.

> --- glib.orig/testglib.c	Wed Mar  1 04:44:09 2000
> +++ glib/testglib.c	Tue Jun 27 12:30:31 2000
> @@ -795,7 +795,37 @@
>    g_string_insert_len (string1, 5, "last", -1);
>    g_assert (strcmp (string1->str, "firstlast") == 0);
>    g_string_free (string1, TRUE);

I've applied this change, but since the canonical location
for 

I've also added g_string_hash() to go along with g_string_equal(),
since these two functions are universally paired in GLib.

guint
g_string_hash (const GString *str)
{
  const gchar *p = str->str;
  gint n = str->len;
  guint h = 0;

  while (n--)
    {
      h = (h << 5) - h + *p;
      p++;
    }

  return h;
}

Regards,
                                        Owen