Re: Performance data for nautilus

[Sven Neumann <sven gimp org>]

Hi,

Alexander Larsson <alexl redhat com> writes:

> Very interesting. As usual it shows up GObject at its worst. I fixed a 
> leak in nautilus and a couple of places that were excessively typesafe, 
> but a lot of this is hard to fix.
> 
> I've attached the interesting data for gtype.c. One thing that strikes me 
> is:
>   * *********************************************************
>   *  Function:                      g_type_check_instance_is_a
>   *  Called:                          1106240 times
>   *  Function time:                 121654830 cycles ( 2.25% of .root.)
>   *  Function+descendants time:     344618647 cycles ( 6.37% of .root.)
>   *  Distribution to Callers:
>   *    203845 times (18.63% of f+d time) g_object_ref
>   *    210290 times (19.22% of f+d time) g_object_unref
>   * *********************************************************
> 
> It seems a full 2.4% of the nautilus execution time is spent making sure 
> that what you pass to g_object_ref/unref() is actually GObjects.
> I'm not sure it's really worth that. Maybe we should change G_IS_OBJECT to 
> g_type_fundamental (((GTypeInstance *)(instance))->g_class->g_type) == 
> G_TYPE_OBJECT. Or maybe we should just totally dump the typecheck?

shouldn't a production system have glib compiled with
G_DISABLE_CAST_CHECKS anyway? I don't see why we should remove checks
that might help developers if they can be easily disabled at compile
time. Actually I have never understood why the extensive cast checks
aren't disabled by default. Perhaps gtk+ and friends should even be
compiled using G_DISABLE_CHECKS.


Salut, Sven