Re: g_strstrip question.

[Tristan Van Berkom <vantr touchtunes com>]

> If the same memory is used and the end is NULL terminated, then you have
> a memory leak.  Even if you free the original string, it will look for
> '\0' and free up to that point.  Without allocating NEW memory, I can
> not see how you could do this?

when I ask the memory manager (MM) to allocate me
BUFSIZ bytes; thats exactly what it does. The MM
has no idea that I'm about to use the allocated
memory for a string.

when you free an adress allocated by the MM; the MM
will look for that address in a table and free the "size"
that was allocated for that address.

the MM doesnt care at all about the values held
in allocated memory (such as '\0').

strchug and strchop are safe because the string _if_
modified will always be smaller than before
(no worries about buffer overflows).

strchop and strchug will also always return its 
argument. (the first character may have changed
value; but its address is the same.)

Cheers,
                -Tristan


> If the same memory is used and the end is NULL terminated, then you have
> a memory leak.  Even if you free the original string, it will look for
> '\0' and free up to that point.  Without allocating NEW memory, I can
> not see how you could do this?



> OK. if they modify the gchar* DIRECTLY then there would have to be
> memory left allocated, unless it just sets the return value as a pointer
> to the first character 'n' along where it feels that is the start of the
> string which satisfies the function.  But if that was true, how would it
> terminate if there were characters to chop on the end? NULL terminator?
> Well then when you freed it, (the original or the new pointer) it would
> not free the WHOLE chunk of memory, it would look for the NULL
> terminator and free up to that.



Martyn Russell wrote:
> Tristan Van Berkom wrote:
>
> > Geez this has gone on long enough.
> >
> > first of all you cant just free any memory address.
> Yes that makes sence.
>
> > char *str = g_strdup(...);
> >
> > --> g_free(str);
> >
> > is good
> >
> > --> g_free(str[1]);
> >
> > is worse than a memory leek (ie. the memory manager
> > never allocated memory for that address)
> Who said anything about freeing from the middle of the gchar *?
>
> > the same goes for the rest of the string.
> > (i.e if the trailing NULL charachter is no
> > longer the last allocated address; the _whole_
> > memory "chunk" will be freed)
> what? assume it is now reduced in size, how does it know where the end
> is, if the trailing NULL character is not the last allocated address?
> and if there is no NULL character, how does it know how big the chunk is
> to free it?
>
> > second of all
> >
> > strchug, strchop, strstrip
> > _all_ modify the string _directly_
> OK. if they modify the gchar* DIRECTLY then there would have to be
> memory left allocated, unless it just sets the return value as a pointer
> to the first character 'n' along where it feels that is the start of the
> string which satisfies the function.  But if that was true, how would it
> terminate if there were characters to chop on the end? NULL terminator?
> Well then when you freed it, (the original or the new pointer) it would
> not free the WHOLE chunk of memory, it would look for the NULL
> terminator and free up to that.
>
> > NO memory has been allocated.
> >
> > if you want a "chopped" duplicate you need to:
> Yes, ofcourse.
>
> > str2 = g_strchop(g_strdup(str1));
> >
> > otherwise str1 is modified.
> If the same memory is used and the end is NULL terminated, then you have
> a memory leak.  Even if you free the original string, it will look for
> '\0' and free up to that point.  Without allocating NEW memory, I can
> not see how you could do this?
>
> > Cheers,
> >                       -Tristan
> >
> >
> > Martyn Russell wrote:
> >
> >
> > > I think you'll find that 'b' is a new piece of memory and so is 'c',
> > > therefore, you were right the first time in stating each variable would
> > > have to be freed.  I am curious about one thing though, if g_strstrip is
> > > a DEFINITION of a combination of g_strchug and g_strchop, that in it's
> > > self should be a memory leak shouldn't it?
> > >
> > >    gchar *a = NULL;
> > >    gchar *b = NULL;
> > >    gchar *c = NULL;
> > >
> > >    a = g_strdup("   mystring   ");
> > >    b = g_strchug(a);
> > >    c = g_strchop(b);
> > >
> > >    /* WHERE c is now the string u want */
> > >
> > >    if(a != NULL) g_free(a);
> > >    if(b != NULL) g_free(b);
> > >    if(c != NULL) g_free(c);
> > >
> > > Martyn
> > >
> > > _______________________________________________
> > > gtk-app-devel-list mailing list
> > > gtk-app-devel-list gnome org
> > > http://mail.gnome.org/mailman/listinfo/gtk-app-devel-list
> > _______________________________________________
> > gtk-app-devel-list mailing list
> > gtk-app-devel-list gnome org
> > http://mail.gnome.org/mailman/listinfo/gtk-app-devel-list