Re: GTK free function doesn't appear to have any affect.

[Allin Cottrell <cottrell wfu edu>]

On Sat, 15 Jun 2013, dE wrote:

> On 06/15/13 14:24, dE wrote:
> > On 06/15/13 03:35, Chris Vine wrote:
> > > On Fri, 14 Jun 2013 23:03:55 +0530
> > > dE <de techno gmail com> wrote:
> > > > On 06/14/13 22:09, Chris Vine wrote:
> > > > > On Fri, 14 Jun 2013 21:41:05 +0530
> > > > > dE<de techno gmail com>  wrote:
> > > > > > On 06/14/13 17:02, Matthias Clasen wrote:
> > > > > > > On Fri, Jun 14, 2013 at 3:27 AM, dE<de techno gmail com>  wrote:
> > > > > > > > I was monitoring the memory usage before and after execution of
> > > > > > > > g_object_unref and gtk_list_store_clear, and it didnt change the
> > > > > > > > memory usage by a bit.
> > > > > > > >
> > > > > > > > Is this normal (am I doing it right?)?
> > > > > > > What are you monitoring, and how ?
> > > > > > >
> > > > > > > It is i normal that freeing memory does not change the resource
> > > > > > > consumption of the process. The freed memory will be available for
> > > > > > > reuse by malloc, but malloc does not immediately return the memory
> > > > > > > to the OS.
> > > > > > No, filled more than 7GB, swap was at ~350 MB, and then I loaded a
> > > > > > small table which would otherwise take less than 10 MB memory, but
> > > > > > the memory usage increased.
> > > > > Can you post the smallest compilable program which demonstrates your
> > > > > problem (run with G_SLICE=always-malloc set), and with particulars
> > > > > of how you are measuring memory usage?  That should identify if you
> > > > > are doing something wrong with how you are handling the memory in
> > > > > your program.
> > > > >
> > > > > Chris
> > > > You can have the whole source code:
> > > >
> > > > http://pastebin.com/4a5DiMsQ
> > > >
> > > > I'd been distributing it around to fix issues.
> > > This isn't going to help I am afraid.
> > >
> > > On some general observations on your earlier questions however:
> > >
> > > A GtkBuilder object is a plain GObject.  It will be freed by calling
> > > g_object_unref() on it.  That will also cause it to release its
> > > references to the objects it has created.  Whether that will destroy
> > > those created objects depends on whether there are any other
> > > references to them which have been acquired, such as by their having
> > > been put in a container.  If so, then you need to release those other
> > > references as well in order to destroy the created objects and free
> > > their memory.  Top level windows need to have gtk_widget_destroy()
> > > called on them.  If you want to remove a widget from a container or top
> > > level window but keep the container or top level window alive, you can
> > > cause the container to release its reference with
> > > gtk_container_remove().  If you want all containers holding a reference
> > > to a particular GtkWidget to release their references so leading to the
> > > widget's destruction, call gtk_widget_destroy() on the widget.
> > >
> > > For plain GObjects, namely those which are not created with a floating
> > > reference (and so are not derived from GInitiallyUnowned/GtkWidget),
> > > to free them you need to have called g_object_unref() on them explicitly
> > > as well as destroy (or remove them from) their container (if any):
> > > except that GtkBuilder will already have done that for you if it
> > > supplies a plain GObject already embedded in a container it has
> > > supplied. For objects derived from GInitiallyUnowned/GtkWidget,
> > > destroying (or removing them from) their container by one of the means
> > > mentioned above is enough.
> > >
> > > You may be aware of all that.  If so I am afraid you need to break your
> > > code down to see whether you have discovered a referencing bug (not
> > > likely but possible - I have reported referencing bugs before now), or
> > > you have neglected to release something somewhere.
> > >
> > > Chris
> >
> > Yes, I realized that over time, but there appears to be something wrong 
> > with g_object_unref or in general all GTK free functions on my system.
> >
> > In this piece of code --
> >
> > #include <stdio.h>
> > #include <gtk/gtk.h>
> > #include <stdlib.h>
> > #include <string.h>
> > #define COLS 300
> > void main (  ) {
> >        gtk_init( NULL, NULL );
> >
> >
> >     int i, j;
> >    char *temp;
> >    GtkTreeIter current;
> >    GType *str_type = g_malloc (sizeof (GType) * COLS);
> >     for (i = 0; i <= COLS; i++)
> >         str_type [i] = G_TYPE_STRING;
> >    GtkListStore *store = gtk_list_store_newv (COLS, str_type);
> >    g_free (str_type);
> >    for (i = 0; i < 300000 ; i++) {
> >       gtk_list_store_append ( store, &current );
> >        for ( j = 0 ; j < COLS ; j++ ) {
> >         gtk_list_store_set ( store , &current , j , "Hello world Hello 
> > world Hello world!", -1 );
> >       }
> >    }
> >    printf ( "freeing after 10 seconds\n" );
> >    sleep (10);
> >    free_ptr ( store );
> >    printf ("Now freed\n");
> >    sleep (10);
> > }
> >
> > free_ptr ( GtkListStore *store ) {
> >        gtk_list_store_clear (store);
> >     g_object_unref( G_OBJECT (store) );
> > }
> >
> > Calling free_ptr ( store ); increases memory usage, even during the last 10 
> > second free period. It'll take at most 7GB of memory.
> >
> > Here there're no GtkWidgets involved. GtkListStore inherits directly from 
> > GObject.
>
> Is this piece of code itself ok?

Certainly not. "void main()" is not valid C for a start (main 
must return int), then the first loop runs off the end of the 
str_type array...

Allin Cottrell